Last posted 7 Nov 2026, 09:06 IST: 121 changelogs checked, 23 new entries.

Track spend by project tags (beta)

Eligible Claude Enterprise orgs can tag projects with codes such as charge codes or client-matter numbers and see spend by tag.

Adobe data store (preview) and location-aware model availability

Preview data store for Adobe Creative Cloud assets, Express templates and Lightroom presets; models not allowed by a project's resource-location policy can no longer be switched on.

Teammates, fully local Computer on Mac and Voice mode

Hand ongoing jobs to teammates, run Computer fully on a Mac, use hands-free Voice mode, and install connectors and skills via Agent Plugins.

Rebuilt Auto Router and new Auto Code router

Auto Router is rebuilt and Auto Code added for coding; both pick a model and reasoning effort per request from the models governance policies allow.

fal multimodal models come to Microsoft Foundry

Microsoft and fal partner to expand video and multimodal generation models available in Microsoft Foundry.

Voice chat with agents

Realtime voice calls with agents in the Agents Window, IDE, web and iOS, powered by Grok Voice; zero data retention with Privacy Mode on.

Copilot impact dashboard and usage API add chat, agent and code-review metrics

Impact dashboard adds Copilot Chat and VS Code Agents window usage; usage API adds code-review addressed rate and flags users with missing telemetry or old client versions.

Get support directly from Agent

Agent checks incidents and docs for support questions and, on Pro, can file a support ticket with debugging details.

Personal Finance in Claude (beta)

Beta that links bank, card, loan and investment accounts so Claude can show balances and spending. Pro, Max and Ultra plans, US only.

New security and admin controls for Claude Enterprise

Enterprise plans get a security center, inference hooks, centralised management and extra access controls, several in beta.

Also announced by Claude Platform API

Vercel Schedules and Security Review in public beta; malicious npm packages blocked

Schedules dispatches cron or one-time tasks to Functions or Queues; Security Review scans every push and Vercel Agent can open fixes; installs now block known-malicious npm versions.

Kiro IDE 1.4.121 (and 1.4.86)

Adds pre/post compaction hooks, workflow run search and a Workflow tab, read-only memory for sub-agents, and warnings for steering docs Kiro could not fully use.

Kiro CLI 3.1.0

Adds PreCompact and PostCompact hooks and shows session credit usage in /usage, the status line and workflow step rows.

Build live apps with Cursor cloud agents

Cloud agents can build, add auth and a Postgres database to, and deploy shareable live apps, with security checks before publishing. All paid plans.

Gemini analyses Google Forms responses

Gemini in the Forms Responses tab runs Python on responses to report trends, cross-question patterns and next steps; Rapid Release from 2 Nov, Scheduled from 19 Nov.

Notion MCP restore-version tool, SDK beta flags and sync health

New notion-restore-version MCP tool, betas option and typed view filters in @notionhq/client 5.31.0, and live sync health on data source reads.

Switch between teams in the Gemini Enterprise mobile app

Users can switch, rename or remove teams in the mobile app without logging out and rescanning the QR code. Generally available.

Excel Copilot links to workbook changes; agents usage report; Cowork uses local Edge

4 Nov notes: Excel Copilot replies link to changed sheets and ranges; admin agents usage report; Copilot Cowork can use the local Edge session; PowerPoint reuses an attached deck's style.

Embed 5 and Rerank 5 available on OpenRouter

Cohere Embed 5 (pro/fast) and Rerank 5 (pro/fast) can now be called through OpenRouter.

Granular access controls for Personal Agents

Site owners can set access actions per Personal Agent; the setting is renamed AI Crawler and Agent access policies.

AI Gateway model aliases route across a pool of models

A model alias is a pool of models called by one name; AI Gateway tries them in order with fallback or uses smart routing.

LiteLLM v1.108.0 (and v1.106.7 patch)

Adds Gemini 4 Argon, a Microsoft 365 Copilot provider and OpenAI Live sessions. v1.106.7 drops the Lens beta from Helm charts: remove lensWorker.enabled before upgrading.

GitHub Copilot in Xcode's coding assistant (public preview)

Copilot works inside Xcode 27.3 beta's coding assistant on macOS Tahoe 26.6+, for all Copilot plans.

Lovable Dreaming memory upkeep, AGENTS.md in subfolders, BambooHR connector (and 8 more)

Lovable tidies project memory in the background at no credit cost, reads AGENTS.md/CLAUDE.md per folder, adds BambooHR, visitor error monitoring and redesigned settings.

Guided Learning in Google Classroom assignments

Teachers can attach a Gemini Guided Learning activity to any assignment; for Education Fundamentals, Standard and Plus.

Langfuse AI Gateway public beta and skill management beta

Apps and coding agents can call models through Langfuse with gateway keys and see cost per user, key and model; a beta adds versioned agent skills tied to traces.

Text watermark on Claude Mythos 5, Opus 4.7 and Sonnet 4.6 output

Text from Claude Mythos 5, Opus 4.7 and Sonnet 4.6 now carries Anthropic's text watermark on listed platforms; Sonnet 4.6 requests with non-default temperature/top_p/top_k may not.

Salesforce federated data store generally available

Gemini Enterprise's Salesforce federated data store is GA, letting users search and act on Salesforce records without ingesting the data.

Knowledge base metadata fields in Sarvam Cowork

Knowledge bases can have custom metadata fields (required, allowed values, filterable) for documents.

ZCode 3.17.2

Adds the ZCode Sites plugin for one-click website deployment, Kimi K3.1 Pro and MiniMax M3.1 Flash model options, an MCP config screen (draft) and several security fixes.

Open models and regional serving in Model Vault

Model Vault and Model Vault Encrypted now run third-party open models on managed dedicated infrastructure, with zero data retention by default.

Browser Run crawl results add URL outcomes and robots.txt diagnostics

/crawl records gain an outcome field (success, robots_disallowed, timeout, page_error) and a job-level robots array for up to 50 origins.

Post-training adds GLM-5.3 Flash and FP4 preparation

GLM-5.3 Flash gets LoRA post-training (2 Nov) and GLM-5.3 Flash / MiniMax M3.1 Flash adapters can now be merged and prepared for FP4 dedicated inference.

Anthropic-compatible Messages API and OpenAI-compatible Responses API

Together AI now serves Messages and Responses endpoints, so apps built on either SDK can switch by changing the key, base URL and model.

GitHub Copilot available in JetBrains Air

Copilot can be selected as an agent in JetBrains Air for local and cloud coding tasks.

BYOK and enterprise custom models in the Copilot usage API

Usage metrics API now reports BYOK and enterprise custom model usage with an is_custom_enterprise_model field; no backfill.

Antigravity app 2.29.0: notebook viewer, Automations, voice mode, remote agents

Adds a native .ipynb viewer and redesigned Automations; voice mode and remote cloud agents roll out gradually to AI Pro and Ultra.

Custom policy message in the Meet Green Room

Admins can show a short custom policy notice before Meet calls, always or only when recording, transcription or Gemini notes are on.

Message threads in beta on the Claude API

Message threads entered beta: the API can keep conversation state on the server, so follow-up requests send only the new messages.

LongCat 2.5 on AI Gateway; Vercel MCP adds CIMD and deployment cards

LongCat 2.5 is available on Vercel AI Gateway; Vercel MCP accepts Client ID Metadata Documents and shows deployments as interactive cards in MCP Apps clients.

Custom avatars from your own photo in Google Vids

Vids users can upload an image to build an AI avatar, refine it and reuse it within a project.

Notion Agent APIs generally available

Agent APIs leave beta, the personal agent works with a personal access token, and PATCH /v1/agents/{id} replaces the deprecated status endpoint; private alpha routes are deprecated.

Lovable: project memory, app imports from Replit/Bolt/v0, SBOM download

Lovable now remembers project and personal context (30 Oct), imports apps from Replit, Bolt, Base44, v0 or Manus (29 Oct) and offers free SBOM downloads (1 Nov).

AgentCore Runtime V2 in five more Regions, including Mumbai

Platform version V2 now runs in Frankfurt, Stockholm, Mumbai, Singapore and Sydney (10 Regions), starting agents from a snapshot for consistent cold starts.

AgentCore Runtime: 250 concurrent connections and OverheadLatency metric for V2 agents

V2 agents can hold up to 250 concurrent connections per session by default, and a new CloudWatch OverheadLatency metric shows the platform's share of latency.

AgentCore Gateway: Web Answer connector GA

Built-in Gateway connector that returns a synthesised answer with citations from one tool call; generally available in us-east-1.

Claude Code 2.1.316 (and 1 earlier build)

New permissions setting blocks writes outside working directories; subagent effort env var; idle compaction over 200K tokens in VS Code/Desktop; gateway pooled budgets and IdP-group upstreams.

MCP integrations for Sarvam voice agents

Voice agents can connect third-party tools over MCP, picked from a catalogue or added by custom server URL.

Kiro CLI 2.37.1 (and 2.37.0)

2.37.0 auto-upgrades V2 agents to V3 when V3 launches (on by default) and adds terminal sign-in for ACP clients; 2.37.1 is a patch.

S&P Global data store (preview); Gemini Notebook Enterprise refresh

Preview data store lets users query S&P Global data their account can access in natural language; Notebook Enterprise gets a new UI, audio overview language choice and custom summaries.

Computer Use generally available in the Gemini API

The Computer Use tool is now GA across the Interactions and GenerateContent APIs.

Sarvam Cowork: custom APIs and 5 GB knowledge-base files

Cowork users can save keys for their own HTTPS APIs for use in personal chats and owned agents (29 Oct); knowledge-base uploads now accept files up to 5 GB (30 Oct).

Browser Run moves to Chrome 152; agents can preview dev servers privately

Browser Run's default browser is now Chrome 152 with a changed Quick Actions User-Agent (30 Oct); Agents SDK browsers can reach dev servers via internal hostnames without public URLs (28 Oct).

OpenRouter guardrails separate organisation limits from workspace defaults

Admins can set org-wide limits and give people different budgets and model access within them; existing guardrails keep working.

Watermarking in Amazon Bedrock documented

New Bedrock guide lists which models add a watermark to their output and explains how to detect it.

OpenCode v2.2.5 (and 1 earlier build)

v2.2.4 adds persistent desktop terminals, opt-in artifacts, guided Claude Agent setup and schedules under opencode serve; v2.2.5 adds worktree/shell shortcuts and a compaction buffer setting.

Replit Cloud Agents for Enterprise; Agent can message other chats

Enterprise Cloud Agents turn a described change into tested code in sandboxes and open a PR/MR; Agent can message or create chats (Pro/Enterprise); GPT-6.1 Terra in Power Mode.

Edit entire presentations with Gemini in Google Slides

One prompt in the Slides side panel can update several slides or a whole deck, restyle it or match another presentation's look.

Usage and Cost Analytics API in beta

New Usage and Cost Analytics API launched in beta; the older Usage and Cost API and several Enterprise Analytics report endpoints are now legacy but still supported.

Codex CLI 0.166.0 (and 2 earlier builds)

GPT-6.1 Terra in model catalogs and codex doctor (0.166); built-in Vertex AI provider, memory v2 on by default (0.165); /presets saved prompts and enterprise MCP auth (0.164).

Call log exports emailed as download links

Voice Agents call-log CSV exports are emailed as download links when ready; exports from the last 7 days show under Recent exports.

LM Studio 0.4.27

LM Link reliability for large transfers, better runtime and ROCm detection, Neural Engine prefill controls on Mac and M5 Metal fixes.

Firewall rules for AI crawlers vs assistants; Vercel Plugin in Antigravity

Managed rulesets can block AI crawler bots and AI assistant bots separately; Vercel Plugin gives Antigravity agents Vercel skills; Strix can use AI Gateway.

Kiro CLI 2.36.0 (and 4 earlier builds)

2.36 adds 'Approve for me' in V3 sessions (auto-approves routine actions, blocks risky ones); 2.35 warns when run as root and applies org MCP and web-tool settings.

Antigravity app 2.28.2: partner plugins in Marketplace

Marketplace adds partner plugins (Figma, Linear, Notion, Stripe), a Featured spotlight and plugin account sign-in; the Conversation History page is removed.

Create, preview and edit files in the Gemini app

Gemini app can now generate, preview and edit files such as slides, spreadsheets, PDFs and code directly within a chat.

Suno Studio updates: Clip Info panel, auto-stretch, fine pitch

Clip settings move to a sidebar with a Clip Info panel; adds auto-stretch for uploaded audio, semitone/cent pitch adjustment, track hiding and grid controls.

Langfuse v4.72.0: AI Gateway analytics and PagerDuty alerts

AI Gateway spans attributed to gateway keys, a gateway analytics tab and onboarding; PagerDuty integration for alerts and prompt changes; trace-level filters in evaluations.

Messages API request size limit raised to 128 MB

Maximum request size for the Messages API and Token Counting API on the first-party Claude API rose from 32 MB to 128 MB.

Gemini Enterprise available in Brazil region (GA, allowlist)

Gemini Enterprise app now available in the Brazil in-country region with at-rest data residency and in-region ML processing, by allowlist only.

Exclude cohort for Voice Agent campaigns

Campaigns can now drop contacts already on them by uploading a CSV with cohort_category set to stop_file, matched by user identifier or phone number.

Five models added to the Cerebras Dedicated catalogue

Kimi K3, GLM 5.3 Flash, Inkling, Reflection AI Beam and MiMo V2.6 Pro join Dedicated; image inputs in tool messages and reasoning controls for Kimi K3 and GLM 5.3 Flash documented.

OpenRouter rebuilds provider routing

Routing now monitors every provider endpoint and balances cost, speed and reliability per request instead of defaulting to the cheapest provider.

AI Gateway virtual models; Particle-1 added

Vercel AI Gateway adds virtual models: custom slugs that save a model's routing, fallbacks and compliance settings. Particle-1 model is also now available.

Kiro IDE 1.4.62 (and 1.4.51)

Windows system-wide installers for x64 and ARM64, drag-and-drop chat attachments, faster reconnect when an Identity Center or Builder ID session expires.

Kiro CLI 2.35.2 (and 2 earlier builds)

2.35.0 warns when run as root/Administrator, makes V3 sessions follow org MCP and web-tool settings, and fixes model fallback and session resume; 2.35.1-2.35.2 are patches.

GitHub Agentic Workflows GA; shared Copilot Automations; GitHub intelligence memory

GitHub Agentic Workflows (natural-language Markdown workflows in Actions) is GA; shared Copilot Automations roll out; memory, context curation and code search help agents keep team context.

Copilot in VS Code Universe releases; plugin defaults and enterprise plugin controls GA

VS Code adds cloud session handoff, standalone Agent Host, side-by-side model comparison and experimental computer use; repos can set default plugins and enterprises require or block them.

Copilot Fusion generally available, with local-model routing preview

Copilot Fusion (successor to HydraFusion) is GA, routing each task across models with Standard, High and Max effort; an experimental preview adds on-device models in CLI, app and VS Code.

Copilot app Universe releases: Projects, computer use, security extension, image generation

Copilot app adds Projects (preview) spanning multiple repos, computer use on macOS/Windows (preview), a code-scanning security extension and image generation/editing (preview, billed separately).

Antigravity app 2.28.1

Adds a conversation navigator, a third-party tools notice in the Marketplace and developer/version details on plugin pages.

Lovable: IdP group mappings, redesigned Share menu, credit run-out alerts

Admins map IdP groups to roles in one place (26 Oct); redesigned Share menu and emails to admins when credits will likely run out before refill (28 Oct).

Zed 1.26.2

ACP config defaults now apply only to new conversations, better fuzzy search in model pickers, OpenCode fixes for DeepSeek and Google models, and optional Pyrefly Python language server.

Immersive explorations in the Gemini app

Gemini adds an interactive library of visual topic explorations that users can drill into layer by layer.

Fill with Gemini in Sheets adds cell-level web citations and 3,000-cell requests

Fill with Gemini now shows the web source behind each generated cell, runs faster and handles up to 3,000 cells per request.

Compact form layout for pivot tables in Google Sheets

Sheets adds a compact pivot table layout that nests row fields in one column and keeps that layout when importing Excel files.

LiteRT in Google Play services: faster GPU inference, smaller apps

LiteRT via Google Play services adds ML Drift-powered GPU inference (about 1.4x faster), a new CompiledModel API and roughly 8 MB smaller apps.

Suno MCP (beta)

MCP clients such as Claude and ChatGPT can create songs, start from uploaded audio and remix library tracks; all plans, uses credits, early-access sign-up.

Claude Code 2.1.312

Adds oidc.user_id_claim for the Claude apps gateway and click-to-send for queued messages; Bedrock/Vertex/Foundry use the shorter system prompt for unrecognised model names; hooks capped at 24 concurrent.

VPC connectivity and static IP egress for custom MCP servers

Custom MCP data stores can route outbound traffic through Agent Gateway in us-central1 and europe-west1, reaching private servers in a VPC or allowlisting by static IP.

Gemini API CLI (gemini-api) beta

A stateless, zero-dependency command-line client for coding agents, scripts and CI that can generate text and media, analyse files and run managed agents via the Interactions API.

Learning agent, admin-built MCP connectors and PowerPoint style references in M365 Copilot

Adds a Learning agent, org-wide custom MCP connectors from the admin centre, usage-based billing for Work IQ APIs, and PowerPoint decks that reuse an attached .pptx style or image.

iCloud Mail and Calendar plugins in ChatGPT

iCloud users can connect Mail and Calendar to ChatGPT to search email, check schedules and edit calendar events.

Custom moderation controls in the xAI API

Chat Completions and Responses requests can include a guidelines array to set how Grok handles the nsfw, violence and profanity categories.

CSAM screening across OpenRouter

OpenRouter will scan inference requests in flight for child sexual abuse material; flagged content may be kept for human review and reported where the law requires.

llama.cpp b11936 (builds since b11912)

Adds MiniCPM-OCR-Jev support (scores OCR text against an image via /v1/systemone) and a --models-memory-max server option; --model or -hf is now required.

Copilot server-to-server authentication (public preview)

Services built with the Copilot SDK can act for an organisation through a GitHub App installation without a signed-in user; usage is billed to the organisation.

Antigravity app 2.28.0 (and 2.27.0)

Adds a built-in browser and PDF viewer in the side panel, any-file attachments up to 20 MB, word-level diff highlighting and Markdown code-block previews.

Lovable workspace events and hooks, Scrive connector, redesigned sidebar

Business/Enterprise admins can receive project events and use hooks to allow or block actions; Scrive e-signature and eID connector; Knowledge renamed Instructions; IdP group mapping in one place.

A2UI Express for faster agent UIs

A2UI Express compiles compact Python-style syntax into standard A2UI JSON, cutting token use and latency for agent-generated UIs with no frontend changes.

Runway Ads

Analyses your top-performing ads on Meta, Google and TikTok, creates on-brand variants for each placement and publishes the ones you approve; included with every paid plan.

Seedance 2.0 bitrate option on Runway API

Seedance 2.0, Fast and Mini accept a bitrate option (standard or high) at the same price; new Workflows guides cover versioned multi-step endpoints.

n8n 2.46.0: personal credentials in team projects

Users can use their own credentials in any team project without sharing; runs are tied to who started them and teammates can't run workflows using credentials they lack.

Claude Code 2.1.311

ultrathink now runs at xhigh effort on Opus 5.5, Sonnet 5.5, Haiku 5.5 and Fable 5.1; self-hosted runner gets idle-hold flags; faster startup for Pro and Max accounts.

Social Technologies MCP lets agents create and configure Meta apps

An AI agent can create and configure Meta developer apps and link Business Portfolios through the Social Technologies MCP server.

Cloudflare Computer v0.7.0: resumable agents and approval waits

Agents SDK package @cloudflare/computer v0.7.0 lets agent code resume after a deploy or crash and pause until an external approval arrives.

Vercel Agent validates alerts; AgentEnv evals and stdin in Sandbox

Vercel Agent can triage alerts before they page anyone; Scale AI AgentEnv evals run on Vercel Sandbox; Sandbox commands now accept stdin, so stdio MCP servers can run there.

Cline v4.1.26 (and builds since 16 Oct)

Introduces Cline Cloud sessions with a Local/Cloud model toggle, GPT-6 support on OpenAI and Azure, a Desktop app with terminal and local-models page; default model moved to Gemini 4 Argon then back to Claude Sonnet 5.5.

Copilot assisted permissions generally available

Assisted mode in Copilot CLI, the Copilot app and VS Code uses a model to auto-approve clearly safe commands, file paths and URLs and asks for approval on the rest.

Notion API: connection health, notion-fetch for connected sources

Developer portal adds a Health tab for connections; notion-fetch can open connector search results; page markdown and old_str matching now ignore pending suggested edits.

Ollama v0.40.11

Adds speech generation on macOS via ollama run, an OpenAI-compatible /v1/audio/speech endpoint, and the d3 decision model family (2B, 4B, 9B, 27B).

Claude Agent SDK (Python) v0.2.174 (and 10 earlier)

Mostly bundled Claude CLI bumps (to 2.1.310); v0.2.166 added typed hook output classes and SystemInitData; v0.2.170 fixed parallel tool-call session messages.

Claude Code 2.1.310 (and 3 earlier builds)

Monitor tool and plugin monitors now work on Bedrock, Vertex and Foundry; Enterprise custom roles for cloud sessions and Remote Control; MCP timeouts in non-interactive runs no longer trigger repeat calls.

ZCode 3.16.3

Copy device ID and login links, per-session document view state, shared plugin process, and fixes for IME input lag and terminal tabs.

Hub: passkeys, OpenUSD 3D previews and external collaborators

Passkeys for sign-in and MFA (21 Oct), interactive 3D previews for OpenUSD and SimReady datasets (22 Oct), and inviting external collaborators from repo settings (23 Oct).

Ollama v0.40.10

Adds Nemotron 3.5 Bolt E4B, a small Gemma 4-based vision-language model with 128K context and a speculative-decoding draft model.

Kiro CLI 2.34.0

Local MCP tools can pause to request input by form or link (MCP elicitation); more workflow controls; memories can be deleted from /memories; new model picker.

Devin Desktop v3.11.212 (and v3.11.216): new Agent Window

New Agent Window matching the Devin web app: up to 8 sessions side by side, Files and Changes views, terminal and preview tabs. Enter now steers a running turn.

Copilot usage metrics API adds per-client totals and custom skill names

The usage metrics API now reports requests and tokens by client tool (e.g. OpenCode) and shows configured names of custom skills, agents, MCP servers and plugins.

Copilot in Teams turns notifications into code; Visual Studio September update

@mention GitHub in Teams threads to have Copilot reproduce bugs and open PRs; Visual Studio gets BYOM preview (breaking change in Agent mode); usage metrics name custom skills and MCP servers.

OpenCode v2.1.4

MCP input forms during tool calls, experimental background compaction, heap-dump debug command and SDK connectClientMcp().

Replit Monitoring Agent, agent-made skills and zero-credit app hosting

Monitoring Agent investigates production issues and validates fixes; Agent suggests workspace skills; chat sharing; Stripe Atlas discount; eligible apps keep running at zero credits up to a limit.

Act on Gemini Notebook notebooks from the Gemini app

Gemini Notebook is now a connected app in the Gemini app, so users can create, edit and query notebooks in chat.

Runway App Studio

Users describe an app in plain language and App Studio builds and hosts it using Runway models, brand assets and files.

Suno iOS app: AirPlay and iPhone Duo support

Suno's iOS app now fits the iPhone Duo screen and adds AirPlay playback for songs and playlists.

Dify v1.18.0

Agents can route workflows via Output Routes, apps move as portable .ifpkg packages, Agent Knowledge Base beta; self-hosters must update all images together.

Claude Models on Google Cloud (Anthropic managed)

Anthropic-operated Claude API reachable through your Google Cloud project, with Google Cloud IAM access control and billing via Google Cloud Marketplace; fast mode for Opus 5.5 is available there.

Claude Code 2.1.309 (and 2 earlier builds)

ultrareview now needs workspace trust (--trust-workspace in scripts); new safety_event OpenTelemetry event; /claude-api api-version migrates code to API 2026-10-21; "@claude review stop" for Code Review.

Single-user quota limits GA in Gemini Enterprise

Admins can cap each user's share of an edition's pooled quota (multiplier per edition and location) on Standard, Plus and Standard Emerging Market editions.

Decisions API generally available with gpt-6-luna

Decisions API leaves beta with gpt-6-luna: shared-input prompt-cache discounts, images in per-question instructions, input_json, and up to 400 questions per request (was 200).

Share profiles within ChatGPT Enterprise workspaces

Enterprise members can share profiles with coworkers in the same workspace; owners control visibility and whether activity appears.

Chat Completions and Responses API generally available

/v2/chat/completions and /v2/responses are GA with GLM-5.3, Gemma 4 31B, DeepSeek V4.1 Flash and sarvam-105b; beta whitelisting no longer needed.

Ollama v0.40.9

Adds Phonon speech generation on Apple Silicon, d3 decision models, and the Flint coding model on Linux and Windows; streaming error-reporting fixes.

llama.cpp b11829 (builds since b11796)

Adds MiniCPM-OCR 1.0 support in multimodal, typed/multimodal input for the rerank endpoint, Nemotron non-standard GQA support and an OpenCL depthwise conv2d.

Bedrock private offers for third-party models can be shared across an organisation

Private offers for third-party models in Amazon Bedrock can be distributed to your AWS organisation from the AWS Marketplace purchase confirmation page.

Kiro IDE 1.4.26

Adds a 'Use workflows' control for how often Kiro starts workflows in a chat, makes installed Powers more reliable, and applies the Memory Off default.

Global default policy for Copilot features; industry comparisons in impact dashboard

Unconfigured GA Copilot features on Business/Enterprise get a global default (rollout to 26 Oct, explicit choices kept); opt-in industry benchmarking in the Copilot impact dashboard.

Antigravity App 2.26.0

Adds a Notifications settings page with sound controls, docks question and permission prompts above the message box, and adds sort and filter to the conversations palette.

OpenCode v2.1.3 (and 2 earlier builds)

v2.1.3 bundles Bun 1.4.4 with security fixes and SDK voice model discovery; v2.1.2 adds free-limit recovery; v2.1.0 brings Windows persistent terminals and breaking provider/compaction changes.

Meet real-time speech translation expands to 63 languages

Gemini-powered speech translation in Google Meet now covers 63 languages with automatic language detection; Rapid Release from 22 Oct, Scheduled Release from 2 Nov.

Google Vault filtered exports for Gemini in Workspace

Admins can filter and export users' Gemini in Workspace conversations by user, OU or date range; available now for Rapid and Scheduled Release domains.

Box and whisker charts in Google Sheets

Sheets creates box and whisker charts natively and keeps them on Excel import/export; Rapid Release from 22 Oct, Scheduled Release from 6 Nov, all users.

Streaming support in Cloud API Gateway (preview)

API Gateway supports long-lived streaming protocols in public preview without spec changes, and gateways get deterministic default hostnames.

Notion API Status properties show default option

Status property schemas now include a read-only default_option field naming the option shown when a page has no explicit Status value.

Claude Duet research preview for Pro and Max

Pro and Max users can try Claude Duet, a new app in which Claude works as a proactive agent (research preview).

Project service tier controls

Projects can be limited to specific service tiers in Platform settings, and organisation settings can restrict tiers for all projects.

Business workspace settings move to ChatGPT Admin Console

ChatGPT Business owners and admins can now manage members and workspace controls in Admin Console; no Enterprise-only features added and seat management unchanged.

Agents SDK createAI replaces workers-ai-provider and ai-gateway-provider

A single createAI function now covers Workers AI, AI Gateway and vendor models; the workers-ai-provider and ai-gateway-provider packages are deprecated.

Ollama v0.40.8 (and 1 earlier build)

v0.40.8 adds Recursive Flint, a 40B MoE coding model with 262K context (macOS first); v0.40.7 adds pplx-decider and fixes Responses streaming tool calls.

llama.cpp b11796 (builds since b11435)

Recent builds let the /v1/messages endpoint accept llama.cpp sampling parameters, add Hexagon NPU weight repacking and OpenCL dp4a GEMM, and fix Metal and SYCL issues.

Claude Fable 5.1, Sonnet 5.5, GPT-5.6 Terra and Luna generally available in Kiro

Four models move to full support in Kiro; GPT-5.6 Sol stays experimental.

Kiro CLI 2.33.1 (and 1 earlier build)

Workspace trust prompt before V3 loads folder config, model fallback in V3 sessions, workflow plan step trees; V3 workflows now start only when asked.

Lovable organisations, auto-created Supabase projects, LlamaParse and n8n connectors

Enterprises can group workspaces into an organisation; Lovable can create Supabase projects; new LlamaParse, n8n and Cube connectors; Seedream 5 Pro and Seedance 2.5 for app AI.

OpenCode v2.1.2

Shows the reset time and recovery options when free messages run out, adds OPENCODE_DISABLE_FREE_MODELS to hide zero-cost models, and improves the installer.

Gemini Notebook adds guided Slide Deck and Video Overview creation

Gemini Notebook adds a step-by-step flow for Slide Decks and Video Overviews with suggested options, optional outline review and a quick-generate path.

Docs Live: draft structured documents by voice in Google Docs mobile

Docs Live rolls out on Google Docs for Android and iOS, letting eligible Workspace users draft structured documents by voice using context from Gmail, Chat and Drive.

Agents CLI extensions for ADK, LangChain and other frameworks

Agents CLI extensions let teams deploy and evaluate LangChain or other agent stacks alongside ADK, and platform teams can add custom policies such as SBOM scans.

User profiles and organisation webhooks in beta; admin web fetch controls

User profiles and organisation webhooks launch in beta, and admins get controls for web fetch and domain restrictions that also apply to Managed Agents web tools.

Inference hooks leave beta for Enterprise

Inference hooks are now out of beta for Claude Enterprise organisations.

Claude Tag API in beta for Claude Enterprise

Beta /v1/organizations/claude_tag/ endpoints let Claude Enterprise organisations read and change Claude Tag settings for Slack channels and workspaces; not available to Console organisations.

Antigravity Enterprise Gateway support

Organisations can route Antigravity model requests through a central gateway for inspection, policy enforcement and gateway-key auth, billed pay-as-you-go.

Locked chats in ChatGPT

Selected conversations can be hidden from history, pinned chats and search, and opened with a 4-digit PIN or device authentication.

ZCode 3.16.2

Multi-folder local projects, persistent remote sessions, plugin marketplace, on-demand tool search and MiMo model support.

Think agents move to @cloudflare/computer workspace

Think and Agent Skills now use the @cloudflare/computer workspace; @cloudflare/shell is deprecated.

Bedrock prompt shortening; grounding checks get Standard tier

Advanced Prompt Optimization can shorten prompts per template and supports application inference profiles; Guardrails contextual grounding checks support the Standard tier.

Kiro IDE 1.4.10: Memory off by default

Kiro IDE 1.4.10 switches the new Memory setting to off by default, makes approved workflow steps follow plan feedback and fixes chat, workflow and MCP issues.

Cursor Agent SDK

Developers can define an agent as a code project, test it with evals and deploy it to Cursor or their own infrastructure.

Antigravity 2.25.0

Antigravity app adds plugin badges on managed sidebar projects and auto-hides empty projects, plus window shortcut fixes.

vLLM and SGLang engines on TPU; Coral NPU open design

Native PyTorch backends for vLLM and SGLang on TPU via TorchTPU; Coral NPU RISC-V design completed with IREE/MLIR compiler v1.0.

Deepgram self-hosted October release; Configure confirmations

Self-hosted release 261020 needs NVIDIA driver 580+ with the open kernel module; Nova-3 Configure messages now get success/failure replies.

n8n 2.45: durable scheduler on by default

Schedule Trigger runs are stored in the database before they fire, so restarts no longer drop runs; legacy scheduler env variables are deprecated. Dated 20 Oct on n8n's changelog.

Claude Code 2.1.304 (and 1 earlier build)

Adds /auto-mode-setup guided auto-mode configuration, a decider:"person" option for mod tool checks and managed-settings ceilings; --worktree no longer deletes worktrees on detached HEAD without asking.

Ultrafast Computer models, Decisions API and connector controls

Ultrafast GPT-6 Astra and GPT-6.1 Sol in Computer for Max, Gemini 4 Argon in Computer for Pro/Max, pplx-decider v1.2 and pplx-embed-v2 models in the API, and role-based connector access for Enterprise.

Cloudflare API MCP server supports step-up authorization

The Cloudflare API MCP server can request extra permissions on demand, so clients can start with fewer OAuth scopes.

Bedrock: Claude structured outputs on both endpoints; retrieval instructions for agentic retrieval

Structured outputs now work for Claude models on bedrock-runtime and bedrock-mantle (18 Oct); agentic retrieval on managed knowledge bases accepts retrieval instructions (19 Oct).

Kiro IDE 1.4.0 adds Memory

Kiro can carry repository conventions, corrections and ruled-out approaches across chat sessions; also fixes Markdown Review find, sign-in Cancel and an Ubuntu 24.04 .deb startup failure.

Bolt: update skills imported from GitHub

Skills imported from GitHub stay linked to their source folder so updates can be pulled in when the skill changes (week of 19-25 Oct).

Copilot model policy API generally available

REST endpoints to read and set Copilot model policies are GA at enterprise, enterprise team, enterprise app and organisation level.

Gemini Notebook in Classroom gets insight summaries for teachers

Teachers get Gemini summaries of how students use assigned notebooks; admins get external sharing controls including trusted domains.

Gemini meeting recaps in Google Chat; publish Gemini apps to Google Sites

After Meet calls with notes, Gemini posts a recap card with decisions and action items in Chat (Scheduled Release from 27 Oct); Gemini-app creations can be exported to a Sites draft.

Gemini in Sheets analyses BigQuery data in natural language; Gemini in Sheets on iOS

Users with query access to a BigQuery-connected sheet can ask Gemini questions and build dashboards over the full dataset; Gemini in Sheets also rolling out on iOS from 15 Oct.

Gemini in Google Sheets on iOS

Gemini in Sheets reaches iOS: users can ask questions about their data, get insights and create charts from their phones.

LiteRT Tensor API for native edge model authoring

New C++ API in LiteRT gives developers direct control of the model graph to run ML computations across edge hardware.

Langfuse experiments show which metric moved and why

Charts every score across experiment runs, summarises results against a baseline and compares candidates item by item by score threshold.

Small serverless rate-limit tier widened to models under 800B

The Small tier now covers models under 800B total parameters (was 600B) and Medium covers 800B to under 1.6T; ceiling values are unchanged.

DeepSeek opens its in-house search API to developers

The search engine behind DeepSeek's web and API search is now available to developers; details are in the Search & Fetch guide.

AI Gateway logs show sessions; Elorian Glimpse added

AI Gateway logs now group requests into sessions with cache hit rate, cost, turns and the coding agent behind each one; Elorian AI's Glimpse model is also available through the gateway.

Claude Code 2.1.302 (and 1 earlier build)

Fork subagents now on by default in -p, Agent SDK, desktop and VS Code; new CLAUDE_CODE_AUTO_MODE_MODEL setting; 2.1.301 adds a usage-limit retry watchdog and LaTeX rendering in VS Code.

Citations for custom MCP data stores; Workflow Builder adds Confluence and Jira Data Center

citation_mappings now shows source citation chips for custom MCP servers (GA); Workflow Builder can search and act in Confluence and Jira Data Center.

Slides in ChatGPT

ChatGPT can now create presentations that teammates co-edit in Space; rolling out to eligible Pro, Business and Enterprise users.

Prompt-to-voice and nine new GPT-Live voices

API users can create custom voices from a text description (free to create; speech billed as GPT-Live sessions). Nine new voices added for German, French, Hindi, Japanese and Korean.

Artifact SDK in hosted sandboxes; Ultrafast mode in Agents API

OpenAI-hosted sandboxes get the Artifact SDK plus skills for spreadsheets, presentations and Word documents; Ultrafast mode is set via agent.service_tier.

Dual-channel call recordings for Voice Agents

Sarvam Voice Agent call recordings are now stereo, with the caller on the left channel and the agent on the right, enabled for all calls.

AI Search keyword search supports Japanese, Chinese and Korean

AI Search keyword search with the trigram tokenizer now matches CJK queries, including unspaced questions; changing the tokenizer triggers a full reindex.

Filter Hub models and datasets by author; faster tensor viewer

Models and Datasets pages can be filtered by author (16 Oct), and the tensor viewer loads large sharded models faster by fetching all shard headers in one request (14 Oct).

Ollama v0.40.6

Adds q2_0 GGUF tensor/file types, launch support for the clef decision model in Ollama's cloud, and a fix for orphaned runner processes on Windows.

AI Gateway adds Perplexity Decider V1.2 and Liquid d1-ultra

Vercel AI Gateway now serves Perplexity Decider and Decider Flash V1.2 (16 Oct) and Liquid AI d1-ultra (15 Oct) through one API key with shared budgets.

Cloud agent canvases and Environments API beta

Cursor cloud agents can build shareable interactive canvases; Environments API in beta, environments support up to 300 repos and load 5x faster.

Copilot in Slack and Teams works across multiple repos

Copilot in Slack and Microsoft Teams can now work across several repositories in one conversation, plus other improvements.

Antigravity 2.24.0 and CLI 1.4.0 (and 3 earlier CLI builds)

App 2.24.0 adds a redesigned side panel, /compact and model choice for automations; CLI 1.3.5-1.4.0 add Auto mode, Claude 5.5 support via ADC and terminal status reporting.

Lovable: publishing policies, Action center, OIDC sign-in, GPT Transcribe

Admins can block or require approval for publishing; Action center for requests and findings; OIDC sign-in for Cloud apps; GPT Transcribe is the default speech-to-text.

OpenCode v2.0.32

Adds /copy for the latest reply or compaction summary, retries Gemini malformed responses, and an experimental voice session SDK API.

Replit Bring Your Own Cloud and Solutions Partners directory

Replit can run in your own cloud tenant (full BYOC private preview; published-app hosting and image exports in early access); new partner directory. Claude Haiku 5.5 added to Power Mode.

Fewer hallucinations on silent multichannel audio

Voice-activity silence trimming is now on for English multichannel transcription, cutting hallucinations during long silent stretches.

Models API reports between_tools thinking support

GET /v1/models now shows whether each model accepts thinking type between_tools, which turns off up-front thinking.

Fast mode for Claude Opus 5.5 leaves research preview

Opus 5.5 fast mode no longer needs access or the fast-mode-2026-02-01 header; every usage tier gets a default fast-mode rate limit. Opus 5 and 4.8 fast mode stay in preview.

Also announced by Microsoft Foundry

Compliance API returns saved Claude instructions (beta)

For Claude Enterprise orgs, the Compliance API can now return members' saved "Instructions for Claude" and "Global instructions" text (beta).

Codex CLI 0.163.0

Searchable /config panel, steering Codex mid-turn with follow-ups, configurable leader key, parallel read-only tools; Daybreak on by default for eligible accounts.

Rosalind Academic Plan; no fixed Sites limit on Plus and Pro

US university labs can apply for up to 30 free ChatGPT Business Standard seats for up to 12 months. Plus/Pro lose the fixed cap on Sites (usage and storage limits remain).

Native Decisions integration in GPT-Live

GPT-Live can call the Decisions API directly or trigger decisions from input/output transcript deltas, so voice apps can act while the user is speaking.

Kiro IDE 1.3.11

Markdown Review reaches all users, plus a What's New browser, clearer workflow notifications and commit-message generation before staging.

Kiro CLI 2.32.0

Folder-wide write approval in V3 sessions, V3 cloud sessions from non-interactive commands, workflows that show their plan first, and Powers/MCP in workflow steps.

Agentic autofix for code scanning alerts GA

Copilot explores the code, proposes a fix, reruns CodeQL to validate it and opens a draft PR for review; now generally available.

Lovable: Sign in with Lovable, growth tools, audit-log export, GPT-6.1 Sol

Sign in with Lovable for apps and any website, OAuth apps, growth-planning tools, SIEM audit-log export for Enterprise, Lovable MCP in Claude via company SSO, and GPT-6.1 Sol for app AI features.

OpenCode v2.0.30

Adds changed-files views by turn, a docked diff panel in the TUI and keyboard shortcuts for session tabs.

Warp 15 Oct: Muse Code as a first-class CLI agent

Warp adds Meta Muse Code as a first-class CLI agent and fixes Claude Code harness completion reporting and cloud agent recovery.

Gmail Live and AI Inbox roll out to paid Workspace

Gemini-powered Gmail Live voice search (Android/iOS) and AI Inbox to-dos and summaries roll out to paid Workspace users, US English only; AI Inbox reaches Scheduled Release domains from 26 Oct.

Runway MCP adds Seedream 5.0 Flash, Fish Audio Drama 3, Mirelo SFX 2.0, BiRefNet v2; project credit caps

Between 9 and 15 Oct Runway MCP gained image, voiceover, sound-effect and background-removal models, and Enterprise admins can cap credits per project.

Collaboration on Suno: shared workspaces and custom models

Shareable workspaces with a shared song library (Pro, Premier) and shareable custom models for collaborators with 2x inputs (Premier).

OpenTelemetry export (beta) for Messages API and Managed Agents

Trace Messages API requests and Managed Agents sessions in your own observability backend; SDK spans in 7 languages, API spans need an OpenTelemetry destination set in the Console.

Managed Agents scheduled deployments on Claude Platform on AWS

Claude Managed Agents scheduled deployments are now available on Claude Platform on AWS.

Claude Code 2.1.300

Adds virtualText for mod prompt decorations, a prompt.change mod hook and /advisor on Claude Platform on AWS, plus many fixes.

Claude Code 2.1.299

Idle compaction for conversations over 200K tokens before cache expiry; MCP allow rules now match only the named server (mcp__a no longer matches a__b); MCP 2026-07-28 negotiated by default for claude.ai connectors.

Gemini 3.8 Flash GA in the UK with in-region processing

Gemini 3.8 Flash is generally available in europe-west2 (London) with in-region data residency and ML processing, on by default in the web app there.

Ollama v0.40.5 (and 1 earlier build)

Adds Nemotron 3.5 Super, cloud access for decision models, clearer cloud error messages and llama.cpp/MLX updates; v0.40.3 was a bug-fix build.

OpenRouter Private Endpoints

Enterprise customers can route their own fine-tuned models and dedicated deployments through OpenRouter alongside the public catalogue.

Kiro CLI 2.31.1

Inline Mermaid diagram rendering in chat; saved settings now apply to headless V3 runs; several fixes.

HydraFusion Max and a bigger HydraFusion model pool

Copilot's HydraFusion now draws on Claude Opus 5.5, GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna; a new HydraFusion Max option targets harder tasks.

OpenCode v2.0.29

Adds remote device pairing via OpenTunnel (QR code or link), provider account switching, and automatic continuation when a response hits the output-token limit.

Zed 1.24.2

Terminal tabs can be dragged into the Agent Panel, custom Amazon Bedrock models get tool use, images and thinking, and long agent threads use less memory and CPU.

MCP support in Apigee hybrid

Apigee hybrid 1.17.0 adds MCP support with the MCP data plane running inside the customer's own Kubernetes cluster.

Notion search filter_properties; MCP backlinks and groups tools

Search accepts filter_properties to return only chosen page properties; Notion MCP adds notion-list-page-backlinks and notion-get-groups tools.

CrewAI 1.15.28 (and 4 earlier builds)

Adds crewai eval reporting where a stopped model broke, --models overlays with cost/time tracking, XPU and DeepInfra providers, Oracle integrations; context windows now from the LiteLLM catalog.

Langfuse Python SDK v5, JS SDK v6 and organisation-level LLM connections

New major SDKs for Langfuse v4 (experiment links, typed metadata filters, compressed span export); LLM credentials can be set once per organisation.

Workflow Builder connects to Microsoft Teams, Notion and Salesforce

Gemini Enterprise Workflow Builder workflows can now search data in and automate actions across Microsoft Teams, Notion and Salesforce.

Antigravity in the Gemini Mac app; Customize menu with 40+ apps

Antigravity coding agents can be started from chat in the Gemini macOS app; a new Customize menu adds skills and connects 40+ apps, rolling out on web, mobile and desktop.

Also announced by Google Workspace Updates

Agents SDK persistent browser tool; API MCP server sends Worker issue events

Agents SDK adds browserTool, a Browser Run browser that keeps tabs, cookies and logins across turns; the Cloudflare API MCP server now pushes signed Worker-issue events to subscribed clients.

Post-training on Together AI

New Post-training product lets you write your own training loop on Together GPUs with SFT, DPO and GRPO methods, billed per GPU-minute.

MAI-Image-2.6 and MAI-Image-2.6-Flash on AI Gateway

Microsoft's MAI-Image-2.6 and 2.6-Flash image generation and editing models are available through Vercel AI Gateway with one key and unified billing.

Kiro IDE 1.3.0

Local command sandbox limiting what agent-run shell commands can access (block/restrict/open network), /workflow-run to start saved Workflows from chat, on-demand MCP tool loading.

Kiro CLI 2.31.0 (and 2 earlier builds)

Local command sandbox and local autonomous mode for all V3 users, on-demand MCP tools; 2.30.0 starts the gradual V3 rollout with model-fallback notices.

Origin App Marketplace

Cursor's Origin integrates with 20+ apps for deployment, CI, code review, security, monitoring and project management; the marketplace accepts public submissions.

Copilot managed-settings enforcement and telemetry user attribution

Admins can require Copilot clients to fetch managed settings at startup, with minimum client versions enforced for governance keys starting with sandbox; client telemetry can now attribute activity to users (12 Oct).

Gemini CLI v0.64.0

Atomic file writes, serialised file tool operations, CLI state recovery from backup, and lower memory/disk use in long sessions.

Lovable WebMCP tools for browser agents, image editing, Loops connector

Apps can expose WebMCP tools that browser agents run in the user's tab; image editing in the Files tab; Loops connector; fixed list of supported Postgres extensions; free-allowance coverage in usage.

OpenCode v2.0.28

Per-project auto-accept for permissions, native Windows ARM64 builds, official WinGet packages, optional Figma plugin with MCP, and 15-minute streamed-chunk timeout for reasoning models.

Skills in Workspace delayed; Gemini app skills roll out to mid-November

Gemini app skills began a gradual rollout on 13 Oct, finishing by mid-November; skills inside Workspace apps are delayed until further notice.

Notion 3.8: Context for your agents

MCP-compatible agents can read and update Notion; Notion agents use context from 150+ MCP tools; Skills API syncs skills with ChatGPT and Claude; GPT-6.1 Sol, Opus 5.5 and Sonnet 5.5 added; admin MCP controls.

Notion MCP can delete pages and databases

Two new Notion MCP tools move up to 20 pages or databases to Trash after user confirmation; they can be restored.

Kling 2.5 Turbo Pro, Kling 3.0 and Kling O3 on Runway Dev

Kling text, image and video generation models (including 4K variants) are available in the Runway API, billed per second.

Langfuse redesigned session view; custom System One decision models

Sessions show messages across traces as one conversation with a navigation sidebar; the System One adapter accepts a custom base URL for decision-model evaluators.

Instant Voice Cloning in the Python and TypeScript SDKs

Instant Voice Cloning (beta) is now exposed in the sarvamai SDKs with one-shot, HTTP streaming and WebSocket calls; needs Python 0.1.37+ or TypeScript 1.1.12+.

ZCode 3.15.1 (and 1 earlier build)

Adds MCP App UI plugin support, per-sub-agent model and tuning options in workflows, and a referral campaign, plus bug fixes.

Agents SDK adds web_fetch and web_search tools; OpenCode 2.0 harness

Agents SDK gains a web_fetch tool (webpages, PDFs and Office docs to Markdown) and a web_search tool, plus an experimental harness to run OpenCode 2.0 in a Durable Object.

Contributor permission for access tokens

Fine-grained and service-account tokens can be allowed to create repositories, with write access only to repos they created.

Ollama v0.40.4 (and 1 earlier build)

Adds decision-model support in the CLI (ollama run --ask), Gemma 4 prefill speed-ups on NVIDIA via MLX runner, multimodal embeddings via llama-server and speculative-decoding fixes.

Devin Desktop v3.10.67

Supports MCP servers on the 2026-07-28 spec, lets enterprise admins lock extensions to one marketplace URL, new folder/git @-mentions and private-preview remote MCPs from Devin Cloud.

Antigravity 2.23.0

Permissions preset picker in the message box, full-screen image/video viewer, .zip and folder attachments, and a 'Fix with Agent' option when an MCP server fails to start.

OpenCode v2.0.27

Built-in worktree tools in Code Mode, confirmation before 'opencode pair --remote' exposes a server, installable web app, and Vertex AI routes for Mistral models.

Gemini next steps on Google Meet post-meeting screen

After a meeting with 'Take notes for me' on, Gemini suggests next steps: draft follow-up emails in Gmail or export action items to Google Tasks.

TTS audio effects, turn-prediction STT; seed field deprecated

TTS and Text to Dialogue accept audio_effects at no extra cost; Realtime STT adds turn_prediction commits; call queueing on by default for new agents; TTS seed and KB auto_discover deprecated.

LangSmith Cloud update (12-19 Oct): tracing-only API keys, ScaleDown gateway provider

Adds tracing-only API keys, a ScaleDown provider in the LLM Gateway and API-based creation of GCP BYOC data planes, plus admin and Engine changes.

LiteLLM v1.105.0

Day-one Claude Sonnet 5.5 and GPT-6.1 Sol support, Decisions API (/v1/decisions, /v1/systemone) across eight providers, new Prism, Sail and Cortecs providers, and litellm.agent(); check breaking changes before upgrading.

Comments on Bolt projects can be sent to Bolt as prompts

Users can pin threaded comments to spots in the project preview, and Editors/Co-owners can send selected comments to Bolt as prompts with visual context; preview theme options added.

OpenCode v1.19.0

Adds 'opencode upgrade --major' and a /update command to move to OpenCode 2 (updates no longer auto-install), allows OpenAI's ultrafast service tier, and hardens PTY connection checks.

Session affinity for Chat Completion V2

Optional session-affinity header on gemma4 at POST /v2/chat/completions lets multi-turn chats reuse cached prompt context for lower latency.

Also announced by Sarvam docs

Cloudflare API MCP server serves Cloudflare skills

The Cloudflare API MCP server now provides Cloudflare skills to MCP clients that support the Skills over MCP extension.

New controls and chat improvements in Copilot for JetBrains

Copilot for JetBrains adds enterprise-managed default models, a Fix action for diagnostics and an MCP auto-startup setting, and drops support for JetBrains IDE 2025.1.

Dynamic workflows for Managed Agents (beta)

Managed Agents can write and run multi-agent workflows in the background, enabled via the multiagent_20261001 config.

Claude Code 2.1.296 (and 2 earlier builds)

Hook onFailure block, higher MCP tool-description limits, MCP 2026-07-28 negotiated by default for claude.ai connectors, subagent autoCompactWindow, Read allow_large option.

Composer predictions in Codex (beta)

Codex suggests your next message after a response; free during beta for Pro users aged 18+, and predictions don't count toward usage limits.

Decision models on the Hub

Decision models get their own tag, icon and Models-page filter; GGUF versions are tagged automatically from llama.cpp metadata.

AI Gateway adds Liquid d1 and Step 5 Preview; agents can buy domains via CLI

Liquid AI d1 and StepFun Step 5 Preview available on AI Gateway; Vercel CLI lets agents purchase domains; Sandbox creation now returns a ready sandbox.

Kiro CLI 2.29.0

Local V3 sessions can remember useful context and reuse it in later chats, with read, write or off controls.

Lovable: workspace publishing domain, cross-chat tasks, 10 new connectors

Enterprise workspaces can publish all apps under their own domain; chats can reference or task other chats; new connectors incl. Zendesk, Neon, Oracle, IBM Db2, Metabase, Loom; invite-only publishing on Free/Pro.

Replit 9 Oct release

Replit Desktop for Windows in private preview (with Microsoft and NVIDIA), sandboxed local builds; send tasks to existing projects from one chat; TikTok Ads via MCP.

Workspace Studio adds logic and Drive search steps

Workspace Studio automations gain nested conditionals, Gmail reply skipping, Meet calendar outputs and a Drive 'Search for a Doc' step.

Notion API: silent page writes and schema-update fixes

Page write endpoints accept a silent notifications mode; data-source schema updates fix option colours and stop clearing omitted descriptions (behaviour change).

n8n 2.42.6 (and 4 earlier builds)

Blocks creating/updating workflows that use deprecated nodes; adds Nano Banana 2.1 and 2 Lite to the Gemini node; Kimi K3 fix in the self-hosted Assistant (2.43.x pre-releases).

Compliance API covers unified Claude experience chats (beta)

For Claude Enterprise, Compliance API chat endpoints now include unified Claude experience chats; existing access keys work. Models API also reports server-tool support (6 Oct).

Compliance API can download Claude Docs as Word files (beta)

Claude Enterprise admins can now pull Claude Docs documents out through the Compliance API as Word (.docx) files, in beta.

Claude Code 2.1.294 (and 7 earlier builds)

Since 2.1.287: Claude Mods for plugins, Haiku 5.5 as default Haiku, local stdio MCP servers default to protocol 2026-07-28, plugin install --marketplace, claude purge.

Gemini 3.8 Flash GA in Singapore with in-region processing

Gemini Enterprise offers Gemini 3.8 Flash in Singapore with in-region data residency and ML processing.

Ultrafast mode for GPT-6.1 Sol

Responses API adds an ultrafast service tier for gpt-6.1-sol to cut time between output tokens, for all API users subject to rate limits; no price stated.

Also announced by OpenAI release notes, OpenAI Codex

Faster steering in Codex; plugin access by role for Enterprise/Edu

Codex in the ChatGPT desktop app reacts sooner to follow-ups during a run; Enterprise/Edu admins with RBAC can set plugin access per role.

LM Studio 0.4.26

Adds /v1/systemone and OpenAI-compatible /v1/decisions endpoints, Bionic coexistence, CUDA 13 on Windows ARM and llama.cpp runtime 2.50.0.

Kiro IDE 1.2.56 (and 1.2.37)

Adds a Workflows feedback link, improves workflow reliability, spec-only commands and hook matching, and fixes chat, terminal and source-control issues.

Copilot code review: organisation billing options and controls

Org owners can bill reviews requested by licensed members to the organisation's cost centre; admins can limit review requests to users with org or enterprise licences.

Antigravity 2.22.0 (and 1 earlier build)

Search across conversation contents, Scheduled Tasks renamed Automations, grouped agent actions, sidebar drag-and-drop and UI extensions in side pane.

AQuA open-sourced; ML Drift for on-device GPU inference

Google open-sources AQuA, a reference agent that diagnoses quality failures of production agents on Google Cloud, and introduces ML Drift for edge GPU inference.

Midjourney testing Thinking Mode; alpha changelog 7 Oct

Midjourney posted 'Testing Thinking Mode' and an alpha changelog; the index page shows titles only.

Suno Studio timing, clip stretching and fade curves

Studio gets more accurate downbeat detection, shift-drag clip time-stretching, curved fades and pop-out plug-in windows.

Browser-use and computer-use SDK classes (beta)

Python and TypeScript SDKs add classes that run the browser/computer-use tool loop, URL and file policies and approval callbacks.

Codex CLI 0.161.0 (and 1 earlier build)

GPT-6.1 Sol becomes the default model (incl. Bedrock catalog); /mcp login from a session; choose mic/speaker for voice; Bedrock multi-agent V2 support.

chat-latest now tracks the newest ChatGPT model

The chat-latest snapshot follows the newest model in ChatGPT for paid business plans; OpenAI recommends GPT-6 models for production.

Serverless rate limits doubled for Small models

Small serverless models (<600B params) now allow 216M total prompt TPM, 54M uncached and 2.16M generated TPM.

Copilot local sandboxing GA; local Ollama models in CLI; AI secret-detection model

Local sandboxing GA in Copilot CLI, app and VS Code at no extra cost; CLI /model discovers local Ollama models; purpose-built model for AI secret alerts.

Antigravity CLI 1.3.1 (and 2 earlier builds)

Default verbosity changes to medium in 1.3.0; /diff gets cursor and file navigation; Windows command sandbox runs without admin rights (1.2.17).

Lovable: workspace credit limits, Drizzle migrations, new connectors

Admins can set credit limits and alerts; Cloud DB schema changes saved as Drizzle migrations; Klaviyo, Excalidraw+, beehiiv, Ramp connectors; Lyria 3 music models.

Warp 2026.10.07

Kiro CLI joins the third-party coding-agent toolbar; unattended CLI agent runs end as blocked instead of hanging; Windows computer-use agents can record video.

Zed 1.23.2

Adds agent idle-thread retention setting, automatic retries when Google models are overloaded, and better ACP plan updates.

Developer Knowledge API and MCP server (public preview)

Public preview of an API and MCP server that lets AI assistants search current docs across Firebase, Google Cloud, Android and other Google products.

Langfuse: API key expiry, decision-model evaluators, usage by project

API keys can have expiry dates; observations can be scored with gpt-6-luna decision models; new organisation usage chart by project with CSV export.

Admin control over Antigravity Boost and Teamwork

Admins can control user access to Antigravity Boost (GA) and Teamwork (Preview) from Gemini Enterprise.

Copilot Search chat, Regenerate/Switch Model, PowerPoint custom skills

Copilot Search gains in-results chat, responses get Try Again and Switch Model, PowerPoint gets custom skills, and Cowork adoption metrics land in Viva Insights.

Audio uploads in ChatGPT

Paid subscribers and workspaces can upload audio files for transcripts, summaries and questions; availability varies by settings, region and model.

AI Gateway: confidence-based decision fallbacks; Glyph Cluster

AI Gateway can escalate a decision to a fallback model when confidence fails a set condition; stealth model Glyph Cluster is free to call for now.

Kimi K3 India cross-Region inference on Bedrock

Kimi K3 is available through the in.moonshotai.kimi-k3 inference profile in Asia Pacific (Mumbai) and Asia Pacific (Hyderabad).

Remote control for local agents

Users can view and reply to agents running on their computer from the Cursor iOS app.

Update IDE to restore agent activity in Copilot usage metrics

IDEs using the Copilot SDK for agent mode left agent activity out of usage metrics; fixed in VS Code 1.139.0+, other IDEs by November 2026.

Gemini CLI v0.63.0

Enables autonomous plan execution in non-interactive mode and bounds tool output and memory in long-running agent loops.

OpenCode v1.18.35

Adds JSON and Markdown formats for agent-readable stats; xAI tool results now carry supported images.

Models API adds model line and thinking-disable flag

GET /v1/models now returns each model's line (e.g. opus) from 1 Oct, and from 5 Oct reports whether a model accepts thinking: {type: "disabled"}.

Self-serve HIPAA BAA for API organisations

Eligible API organisation admins can accept the standard Business Associate Agreement and turn on HIPAA support from Organization settings.

Also announced by OpenAI release notes

Automations, inline visualisations and new models

Perplexity adds Automations for ongoing work, interactive visualisations in threads, GPT-6.1 Sol, Claude Opus 5.5, MiniMax H3, Seedance 2.5 and Amex business Skills.

P(doom) field on user profiles

Users can add an estimated probability of AI existential catastrophe to their profile, feeding an anonymised community survey.

llama.cpp v0.6.0 (and builds to b11435)

v0.6.0 adds GLM-5.3-Flash, Clef, Ling 3.0 VL and Nimble support, a /v1/systemone decision-model API, llama_batch_ext and a web UI overhaul.

Kiro CLI 2.28.0 (and 1 earlier build)

Classic users can switch to 3.0 from an interactive session, and V3 hooks can match tools by category.

Stock charts in Google Sheets

Sheets adds High-Low-Close and Open-High-Low-Close stock charts, which now survive Excel imports.

Native Markdown (.md) files in Drive and Docs

Open, edit and co-author .md files in Google Docs without converting, with rendered previews in Drive; rolling out from 5 Oct to all Workspace and personal accounts.

Albums

Creators can publish a collection of their work as an album, which notifies their followers.

New and improved Extract (Doc Agents)

Extract is more accurate on Indic, English and mixed-language documents, with lower latency, longer documents, per-value citations and higher rate limits.

Federated query mode for data connectors (Preview)

Gemini Enterprise can query AlloyDB, BigQuery, Cloud SQL and Spanner data in place without ingestion; Knowledge Catalog integration.

Finances expands to Free and Go users (US)

Financial account connection for spending insights and investment tracking now available to Free and Go users in the US on web, iOS and Android.

Also announced by OpenAI release notes

Web Search API (beta)

AI agents can search the web through Ceramic.ai, Exa or Linkup with zero data retention, billed through AI Gateway credits.

Pi Durable harness on the Agents SDK (beta)

Run Earendil's Pi framework as a long-running agent harness with durable persistence on Cloudflare.

Higher serverless rate limits for Small models

Adaptive rate-limit ceilings raised for serverless models under 600B parameters.

Provisioned throughput adds GLM-5.3 and DeepSeek-V4.1-Flash; dynamic rate limits retired

GLM-5.3, GLM-5.3-Flash and DeepSeek-V4.1-Flash available on provisioned throughput; serverless no longer applies traffic-scaled dynamic rate limits.

Copilot code review API support; Balanced default effort

Request Copilot code review through REST and GraphQL with configurable effort; Balanced is now the default level.

Lovable: EU AI data residency, API-key IP allowlist, GPT Live voice

Enterprise EU inference option, IP allowlisting for workspace API keys, announcement banners, GPT Live voice in TanStack Start apps, Lovable MCP server with Meta Muse.

Replit 2 Oct release

Jev model via Replit integrations, GPT-6.1 Sol and Claude Sonnet 5.5 added, Settings now a full page, workspace-level configuration controls.

Change Nova-3 keyterms mid-stream

Replace vocabulary during a live /v1/listen session with Configure messages, no reconnect; Flux STT numerals can also be toggled mid-stream.

Scan notes and documents with ChatGPT camera

The iOS camera can scan multi-page documents and compile them into a single PDF automatically.

Also announced by OpenAI release notes

New ways to shop in ChatGPT

Virtual try-on for clothing and accessories, plus saving products to favourites folders, on web and mobile.

Also announced by OpenAI release notes

Voice Cloning streaming

Voice cloning now streams audio over a single HTTP stream or a persistent WebSocket that accepts incremental text.

Also announced by Sarvam docs

Workers OAuth Provider v1 with MCP 2026-07-28 support

Separate authorization and resource servers, MCP 2026-07-28 compatibility and service-binding token validation.

Code sandbox SDK and CLI

together-sandbox SDK (Python, TypeScript) and CLI run commands in isolated Docker environments using Together API keys; replaces the legacy CodeSandbox integration.

Kiro CLI 2.27.0 (and 2 earlier builds)

V3 gets multi-step Workflows, workflow delegation controls, live steering context and saved prompts as slash commands; Classic interface shows deprecation guidance.

Dynamic workflows in Copilot CLI, app and SDK

Define orchestrated multi-agent processes in code.

Copilot in VS Code: September 2026 releases

VS Code 1.136 to 1.140 add agent automations, agent merge for pull requests and better session management.

Copilot computer use in public preview

Copilot CLI and the Copilot app can control desktop applications on macOS and Windows.

Gemini student hub for Workspace for Education

Central place for study tools such as notebooks and flashcards, now available to Workspace for Education users.

Meta Model API: Muse Code GA, Muse Image API, Muse Glimmer (undated)

Undated items on the hub: Muse Code leaves beta with new plans, Muse Image on the API at $0.01 per image, and Muse Glimmer open-weight model for local agents.

Voice Agents: forwarded call recordings and call traces (undated, Oct 2026)

The Analytics API can download a transferred call's forwarded leg, show a step-by-step event trace, and match calls to telephony provider records.

Instant Voice Cloning: language_code now optional

Sarvam Instant Voice Cloning no longer needs language_code; the language is detected from the text. Pass it when exact pronunciation matters. Undated, October 2026.

Agent session history with Mine/Team toggle (undated, Oct 2026)

Sarvam agent pages add a Mine/Team toggle to view your own or teammates' recent sessions; finished sessions open as read-only transcripts.

Batch API supports Kimi K3.1 Pro and Kimi K3

Batch API accepts kimi-k3.1-pro and kimi-k3 on CN and Global platforms, with cache-write fees and best-effort cache hits. Undated, October 2026.

AgentCore: session endpoints, MCP tasks for long-running tools, larger env vars

Runtime session endpoints (HTTP, streaming, gRPC, WebSocket), Gateway MCP task support for long-running tool calls, and 4 KB environment variables for V2 agents.

AgentCore Runtime: scope access to a session or runtime

Session context keys let sessions sharing an execution role each be limited to their own data with a single policy (platform V2 runtimes). Undated within October.

AgentCore Runtime raises environment variable limit for V2 agents

V2 agents can now use up to 4 KB of environment variables, up from 1.5 KB (direct code) and 2.5 KB (container).

AgentCore Runtime FUSE file-system mounts; Agent Registry in four more Regions

Agent containers can mount storage via FUSE and use it as local files; AWS Agent Registry expands to US East (Ohio), Frankfurt, London and Canada (Central). Undated within October.

AgentCore Memory: time-to-live for long-term memory records

Long-term memory records can be set to expire (minimum 31 days); setting or removing a TTL needs the new UpdateMemoryRecordExpirySpecification IAM permission.

AgentCore Gateway: private CA support for targets

TLS certificates from private certificate authorities on MCP server, OpenAPI and HTTP proxy targets, allowing direct connections to VPC endpoints.

AgentCore Evaluations: failure analysis insights

New Builtin.FailureAnalysis evaluator finds failures across agent sessions, identifies root causes and groups them into insights with suggested fixes.

AgentCore CLI 1.0, Gateway model fallback, InvokeModel and Step Functions connector

AgentCore CLI v1 (agentcore migrate for v0 projects); Gateway adds ordered model fallback, Bedrock InvokeModel translation on /inference, and a Step Functions MCP connector.

Admin API organisation methods leave beta

In the latest SDKs and ant CLI 1.37.0, organisation Admin API methods move to client.organization; the beta namespace stays available.

Claude Code 2.1.286 (and 5 earlier builds)

Sonnet 5.5 becomes default Sonnet (2.1.284); claude --desktop and install-time plugin config (2.1.285); deniedModels managed setting (2.1.283); Bedrock assume_role and MCP URL elicitation (2.1.281).

Kiro IDE 1.2.4

Background multi-agent workflows, safer handling of untrusted workspace configs and enterprise sign-in controls.

HydraFusion in VS Code and the Copilot app

The research preview that orchestrates multiple models now extends to VS Code and the Copilot app.

Antigravity 2.19.1 (and CLI 1.2.10 to 1.2.13)

Message subagents directly, export rendered Markdown as PDF, conversation-only undo; CLI adds --effort and server-requested retry delays.

Zed 1.22.0

BYOK for GPT-6.1 Sol, per-subagent model selection, automatic subagent compaction and a unified/split diff toggle.

Skills in the Gemini app and Workspace (successor to Gems)

Reusable custom instructions called Skills arrive in Gemini and Workspace, replacing Gems.

Also announced by Gemini app

Comments and suggestions in Docs, Sheets and Slides APIs

Developers can programmatically manage comments and suggested edits across the Docs, Sheets and Slides APIs.

SIEM event for meeting-notes consent setting

The SIEM/DLP stream emits workspace.settings.meeting_notes_consent_setting_updated when admins change consent for AI meeting-note transcription.

Flux TTS inline pause and pronunciation controls

Inline pause markers (500 to 3000 ms) and IPA pronunciation controls in batch and streaming modes.

Codex CLI 0.159.2 (and 3 earlier builds)

GPT-6.1 Sol default model incl. Bedrock catalog (0.159.1); opt-in instant interrupt (0.159.0); MCP OAuth client secrets and secure WebSockets (0.158.0).

Sign in with ChatGPT and use your plan allowance in other apps

Single sign-on for participating products, plus a limited preview letting Plus/Pro users authorise apps to draw on their ChatGPT allowance.

Also announced by Devin Desktop

'dot' always-on agents, Meetings plugin and event-triggered automations

Always-on agents on GPT-6 Astra for eligible Pro/Business Premium users; beta meeting notes plugin; automations can start from connected-app events.

Codex Cloud, Codex code review and Codex Security Cloud

Cloud environments for continuous Codex tasks across devices, a Code Review plugin for pull requests, and scheduled GitHub repository scanning with proposed fixes.

ChatGPT Pages, Space and Sites updates

Turn chats into collaborative Pages, organise work in ChatGPT Space (Pro/Business/Enterprise), and edit or schedule published Sites.

Computer use in Agents API; Ultrafast tier for GPT-6 Astra

Agents can now work in OpenAI-hosted browsers (site access, sign-in handling). New Ultrafast service tier cuts time between output tokens for GPT-6 Astra.

ZCode 3.14.4 (and 3 earlier builds)

Dynamic workflows orchestrating sub-agents and Office/Coding mode switch (3.14.0); adjustable workflow concurrency (3.14.3); CAPTCHA removed for free tier.

AI Gateway User Insights task analysis

AI Gateway groups conversations by task to show model overuse and possible savings.

Ollama 0.35.1 (and 3 earlier builds)

Decision models via /v1/systemone (0.35.0), up to ten web searches per response, MLX default on Apple Silicon in 0.40.0-rc0.

Fine-tuning: LoRA rank to 128, longer Qwen 27B context; batch files kept 7 days

LoRA rank up to 128 (from 64); Qwen 27B SFT context 131K and DPO 65K; batch job input, output and error files retained for 7 days.

Official Ruby SDK for MCP reaches Tier 1

Ruby becomes the sixth Tier 1 MCP SDK; mcp gem 1.2.0 fully supports the 2026-07-28 spec.

Resource-level access controls; new data stores and M365 actions

Per-app and per-data-store IAM permissions; new OneDrive/Outlook/SharePoint/Teams actions and several new data stores in preview.

Personalised summaries of health data

Pick charts or metrics in the Health tab to get explanations based on connected health information.

Also announced by OpenAI release notes

Antigravity 2.18.1 (and 4 earlier builds)

Plugin install/management (2.18.1), /plan before building (2.17.0), WSL and Office attachments (2.16.0), Windows sandbox (2.15.1).

Claude Agent SDK (Python) v0.2.161 (and 4 earlier)

New verbatim_prompts option to stop untrusted text triggering file reads or commands (0.2.158); bundled CLI updated to 2.1.284.

crewAI 1.15.23

Adds native Gemini 3.8 Flash support, AMP evaluation of the last traced run in crewai eval and easier platform integration setup, plus tracing and Bedrock fixes.

LangSmith Cloud updates (21 Sep - 5 Oct)

Adds a query_runs tool to LangSmith MCP, wider OpenAI and Vertex cost tracking, Claude Opus 5.5 in Playground and Sonnet 5.5 in LangSmith Chat.

LiteLLM v1.103.0

Unified config precedence, Auto Router capability classifier, MCP allowlisting and OAuth delegation, plus 408 new model catalogue entries.

Dubbing API and Voice Cloning API

New APIs to localise audio/video while keeping each speaker's voice, and to synthesise speech in a cloned voice from a short sample.

Also announced by Sarvam docs

Build plugins for Claude

Developers can submit plugins for Claude through a developer portal with review tracking and usage analytics.

Fix for GPT-6 Sol and Luna image understanding

An image-encoding bug that degraded visual tasks, including computer use, in GPT-6 Sol and Luna has been fixed.

Also announced by OpenAI release notes

safety_identifier request field

Optional field attributes policy violations to end users rather than the API key.

Live resource usage on Jobs

Job pages show a Resources panel with live CPU, RAM, GPU and VRAM usage while a job runs.

Ollama v0.40.0

On Apple Silicon, supported models now run on MLX by default; adds gemma4, qwen3.6, qwen3.5, decision models on MLX and embeddinggemma-2. Page dates it 25 Sep.

Agentic autofix uses Copilot Memory; Slack/Teams updates

Security autofix learns from stored memory and past fixes; Copilot for Slack/Teams gets per-message model choice and duplicate-issue checks.

Replit 25 Sep release

Meta device app building, Muse integration, inline data charts, Airwallex payment testing via MCP, and enterprise access-request workflows.

Meta Connect 2026 developer recap

Meta VR Glasses unveiled with v207 SDK, three build paths for AI glasses, early access to Horizon Create and Horizon Studio, and a VR Start developer competition.

External access controls for ChatGPT Business and Enterprise

Admins can control, from the Admin Console, which sites access connected apps and which apps access ChatGPT Ads.

Hugging Face on Google Cloud Marketplace

Organisations can pay for Inference Providers, Endpoints, Spaces, Jobs and seats via Google Cloud invoices; Jobs now show live CPU/GPU usage.

TogetherLink beta; Tev1-4B-experimental

TogetherLink routes coding agents to Together models with cost tracking; Tev1-4B-experimental serverless at $0.042 input, free output.

Foundry: wider model choice, voice agents, continuous optimisation

Microsoft Foundry adds more model options, voice-agent support and automated agent performance tuning.

Kiro IDE 1.1.70

Configurable terminal timeouts for agent commands and Markdown links to workspace files.

Cloud API Gateway as an MCP server

Google Cloud API Gateway can expose REST APIs as MCP tools through OpenAPI annotations.

Edit updates, thumbnail previews

New options to edit posts and preview image thumbnails.

Cache diagnostics generally available

Cache diagnostics out of beta on the Claude API: no beta header needed and the diagnostics field is always returned. Compliance API local sessions now cover Claude in Chrome (beta, 18 Sep).

Excel Copilot change links and attribution; licence request page

Copilot responses link straight to the Excel cells it changed and Show Changes marks AI edits; admins get a Copilot licence-requests page.

Plugins in Voice conversations

Plugins now work in Voice on web, iOS and Android, and Voice is available in ChatGPT Work.

Sarvam agent workspace updates

Live voice input up to 5 minutes, one-command Slack connection, file sharing by link, artifact comments and faster batched agent outputs.

Kiro CLI 2.24.0 (and 2 earlier builds)

Session-wide tool approval and environment isolation (2.24.0); reasoning/effort controls in /model and cloud sessions (2.23.0).

Rollouts and Security Review

Two new bots: Rollouts tracks changes across environments; Security Review flags exploitable vulnerabilities in pull requests.

Local sandboxing in the Copilot app; configurable code reviews

Sandbox policies limit agent access to files, network and credentials; new code-review settings page and enterprise effort defaults.

Lovable: Claude models without API keys; new cloud connectors

App AI features can use Claude models without keys (GPT-6 Sol/Luna added 25 Sep); AWS, Azure, Google Cloud and Looker connectors.

Zed 1.21.0

SuperGrok sign-in, idle-sleep prevention while agents run, and BYOK for Claude Opus 5.5 and GPT-6 models.

Ground Google Docs prompts on Gemini Notebook sources

Docs can reference Gemini Notebook sources to ground AI drafting with inline citations.

Antigravity SDK supports local models

Offline agent workflows with local models such as Gemma 4 26B via LiteRT and OpenAI-compatible servers.

Runway in DaVinci Resolve; Runway MCP on Cursor and Grok

DaVinci Resolve plugin; Runway MCP on Cursor Marketplace and Grok Bot; Brand Kits and four Kling models via MCP.

Flashcards in ChatGPT

Interactive flashcard creation and practice on all plans.

Keyterm prompting on streaming speech-to-text

keyterms parameter now works on the streaming WebSocket STT endpoints with Saaras v4.

Also announced by Sarvam docs

Temporary rate-limit increase for Qwen 3.8 27B

Developer-tier limit for Qwen 3.8 27B raised from 450K to 750K tokens per minute for now.

Notion file uploads accept CAD/3D; higher MCP limits

File Upload API accepts .dwg, .dxf, .step, .stl and similar; MCP search/query limits raised to 20 calls per 10 s.

Jev as a judge

TypeSafe's Jev decision model can be used as a lower-cost evaluator judge.

Privacy Center; credit scores in Finances (US)

Central privacy-controls hub for consumer plans; US Plus/Pro users can link Experian credit report and VantageScore 3.0.

Effort mode, GPT-6 Astra and Skills marketplace

Perplexity Computer gains selectable effort levels, GPT-6 Astra, a skills marketplace, hybrid cloud/local compute on Mac and Side Chat.

Bolt training opt-out setting

New account setting to stop projects and chats being used for model training; default depends on region.

LLM Gateway adds Nemotron, DeepSeek V4.1 Flash, GLM 5.3 and 5.3 Flash

Four more models available through AssemblyAI's LLM Gateway.

n8n 2.40: agents callable from workflows and Slack

n8n 2.40 introduces agents built in n8n that can be called from workflows, Slack and other channels.

LM Studio 0.4.25

Adds the Splash engine on Apple Silicon M3 or newer with macOS 26.4+.

ChatGPT for Word; multiple accounts per plugin

ChatGPT inside Microsoft Word for drafting, summarising and revising on all plans; several Gmail/Calendar/Contacts accounts can be connected.

Tests for voice agents

Voice agents can be regression-tested with test cases, with API support for managing tests.

Enhance Frame Rate on Runway Dev

Convert videos to 24-120 fps; billed at 1 credit per 2 seconds.

Suno Studio MIDI improvements

More accurate MIDI transcription and new audio stems from a MIDI clip.

Deepgram India endpoint generally available

api.in.deepgram.com keeps data processing within India.

Kimi Web Search API; billing split change (day not given)

Web Search API added at /v1/tools/search; charges now deducted 50% cash and 50% vouchers. Page gives only the month.