AI.info
AI Tools Updates — Page 56
What changed in the AI tools this week — dated, from the makers’ own changelogs.
In inglese
- Arize AX — Connect Perplexity and Cloudflare as AI providers
Perplexity and Cloudflare are now available as AI providers in Arize AX; Cloudflare is also available as a Decision Judge provider.
- Arize AX — Restrict Alyx to one AI provider and model
Space admins can lock Alyx to a single AI provider and model from Space Settings.
- Arize AX — Run trace and session evaluators on datasets
Trace-level and session-level evaluators can now run over a dataset and write one verdict per unit instead of scoring each row on its own.
- Subframe — Edit components across pages
Multi-select the same component across different pages on a canvas and change their properties all at once from the inspector panel.
- Subframe — New keyboard shortcuts
Press ⌘⇧C (Ctrl+Shift+C on Windows) to copy the selected layer as a PNG to your clipboard.
- Subframe — Upload images over MCP
Your agent can upload your screenshots, photos and logos to a project and use them in designs.
- Deepgram — Deepgram Self-Hosted October 2026 Release (261020)
Container images released: API 1.192.55, Engine 3.140.1, license proxy 1.11.2, and billing 1.15.1.
- Scribe — Scribe browser extension 104.34.0
The side panel recommends Scribes for the current page, and onboarding check-ins are tailored to your plan.
- Factory Droids — CLI v0.245.0, Desktop v0.202.0: Chrome cookie import, folder links in the Files panel, voice dictation answers, and CLI and app fixes
The built-in browser can import Chrome cookies on macOS, and clicking a folder path in chat opens that folder in the Files panel.
- LiteLLM — LiteLLM v1.105.2 fixes count_tokens, bridged tool calls and thinking blocks
Chat completions bridged to /responses return text and tool calls in one choices[0].
- LiteLLM — LiteLLM v1.103.6 fixes realtime handshake errors and double spend logs
A refused upstream realtime handshake now sends the client an error event and closes with 1008, 1013 or 1011.
- LiteLLM — LiteLLM v1.106.1 fixes count_tokens, MCP tool calls and thinking blocks
Bare MCP tool names on a single-server scope now go to that server; token count responses carry litellm_count_source.
- Fish Audio — More normalization languages
Normalization adds Turkish and Tagalog rules, Brazilian Portuguese support, and Traditional Chinese detection.
- Slack AI — Security Guidance
This release includes significant security improvements. Updating is strongly recommended.
- Tonic Fabricate — From the plan details, you can now select the mode and models to use when you execute the plan.
Select the mode and models to use when you execute a plan; API keys show who created each key; several server connections support tunnels.
- Tonic Fabricate — Bug fixes and other internal enhancements.
- Omni — October 12, 2026
- Typefully — Address Feedback with Typefully's Assistant
Typefully’s Writing Assistant can address draft comments, apply edits, resolve addressed comments, and undo changes.
- Fern Docs — Role-based access control in local preview
fern docs dev now applies role-based access control the same way a preview link does.
- Splice Create — Get a free experimental synth pad for Splice INSTRUMENT
A free experimental synth pad for Splice INSTRUMENT is available during the drop window and can be kept forever.
- Tines Workflow Automation — 🪐 New Inception connector and AI provider
Added an Inception connector and AI provider.
- Tines Workflow Automation — 💧 New Liquid AI connector
Added a Liquid AI connector.
- Tines Workflow Automation — ✏️ Rename tunnel networks via the networking settings page
Added the ability to rename tunnel networks via the networking settings page.
- Jan — Jan v0.8.7: Shell Switch, Image Size Limit, Project Threads & Agent Subagents
Adds a switch to turn the shell tool off in chat, an image size limit, and built-in subagents to the Jan agent.
- Devin — October 19, 2026
Devin Oncall is generally available to all organizations, and the work log shows current actions without internal housekeeping.
- Agent Studio — From conversation to payment: Sierra and Stripe
Sierra and Stripe let customers pay and get a receipt on the same call.
- Splice Sounds — Get a free experimental synth pad for Splice INSTRUMENT
A free experimental synth pad is available for Splice INSTRUMENT during the drop window.
- Gemini CLI — Release v0.66.0-preview.0
- Gemini CLI — Release v0.65.0
Fixes untrusted-folder workspace settings, resumed-session responses and history, OAuth callback validation, and Ctrl+O terminal resets.
- NovelAI — [Text Generation Notice] Retiring Genji, Krake, and Snek models
Genji, Krake, and Snek are being removed from novelai.net.
- Devin Desktop — October 19, 2026
The work log shows the current step sooner, and Devin Oncall is now available to all organizations.
- GLM Coding Plan — Release v3.16.2
Local projects support selecting multiple folders; remote sessions support persistent processes; added a plugin marketplace and tool search.
- Browserbase — Introducing Stagehand 4.3: OpenAI and Claude SDK integrations, page waits, and more
Use Stagehand with the official OpenAI and Anthropic SDKs; use page.waitForFunction() to wait for application state, updated text, or another browser-side condition.
- Plus AI — Announcing the new Plus AI MCP
The Plus AI MCP allows you to create and edit presentations right from Claude and ChatGPT.
- Stagehand — Introducing Stagehand 4.3: OpenAI and Claude SDK integrations, page waits, and more
Use the official OpenAI and Anthropic SDKs, wait for page state, connect through authenticated CDP gateways, and get the extension as ZIP bytes.
- Mistral AI — Voxtral 3
Voxtral 3 is a voice agent SDK for real-time speech in 50+ languages, accurate transcription, emotional TTS, and tool use.
- OpenHands Cloud — openhands: v0.84.0
Added Azure AI Foundry configuration, Keycloak database backups, and OAuth V2 login; bumped image tags.
- OpenHands Cloud — openhands: v0.85.0
Bumped automation and OpenHands image tags and fixed managed-key test organization setup.
- EV Management System (EVMS) — North Carolina's Electric Cooperatives Launches EVConnectNC, Powered by WeaveGrid, to Maximize Charging During Lower Demand and Power Cost Periods
Launches EVConnectNC, powered by WeaveGrid, to maximize charging during lower demand and power cost periods.
- Squibler — October 20th, 2026
Built a writing agent for longer books and screenplays; updated AI models; improved recovery of interrupted AI tasks and highlight visibility; fixed AI Story Details.