AI TrackerAI news,
sorted for builders.
Launches, price changes and retirements from every AI provider, summarized. The same feed as the app's News tab.
Vercel Connect adds managed Stripe API keys for each agent
Vercel says Vercel Connect now supports managed Stripe API keys per agent. The keys are scoped to the permissions the user picks, no secret key has to be pasted, and keys can be rotated without redeploying.OpenAI explains how it scaled Habitat, its storage platform, on Azure Cosmos DB
In the second part of an engineering series dated October 23, 2026, OpenAI describes Habitat, the online storage platform behind ChatGPT, the API and Codex. OpenAI says Habitat serves over 1 billion users across 40 regions and holds more than 500 petabytes of data, and that its summary of the post claims 12× faster reads for ChatGPT.Kimi K3.1 Pro starts rolling out on Kimi web and mobile
Moonshot's Kimi K3.1 Pro is appearing in the Kimi web and mobile apps, according to TestingCatalog, which posted a screenshot where the app describes it as "Complex tasks and professional work, all-around flagship." Other X accounts (LuminaBench, AiBattle_, ai_for_success) report the same, and say K3.1 Pro and a K3.1 Pro Highspeed variant are also in the Kimi Code CLI. No official Moonshot post was seen when these reports were made.
Cursor agents can now control your computer
Cursor says its agents can now control the computer and drive desktop apps, such as the iOS Simulator, to test the features they build. Cursor's post includes a demo video; it gives no details on availability, plans or limits.Polylane launches proactive agents that watch and fix Cloudflare Workers apps
Boris Tane announced Polylane for Cloudflare, a service that runs proactive agents which continuously watch a Cloudflare account's Workers, Durable Objects, Workflows and similar resources. According to the announcement, the agents fix bugs, improve performance and handle alerts.Vercel explains how Vercel Agent gets access to production
Vercel published a post on how Vercel Agent accesses production deployments. According to Vercel, the agent investigates a problem with read-only access, then proposes a plan together with the permissions needed to carry out the fix, so that asking it to fix a deployment does not hand it broad credentials.Google releases Gemini 4 Argon, its latest frontier model
Google released Gemini 4 Argon on October 13, 2026, and says it is available globally. Google's blog post was shared on Hacker News on October 20. The Gemini API changelog lists gemini-4-argon as generally available from October 13, describing it as Google's most capable model, built for complex coding, agentic, math and science tasks.Codex 0.164.0 adds /presets, inline /model and /permissions arguments and enterprise MCP sign-in
Codex CLI 0.164.0, released October 22, 2026, per its release notes:Vercel AI Gateway adds tako_search, a structured-data search tool
Vercel announced that Tako's structured-data search is now available as a tool on AI Gateway. Instead of pages of text, the tool returns data as JSON with units, sources and dates: Vercel lists stocks, polls, weather and government contracts, and says the data is licensed, cited and up to date.
Moonshot teases Kimi K3.1 in a video; the model is not announced
Moonshot's Kimi account posted a 24-second video without a caption. Accounts that quoted it describe the Kimi Code terminal showing "Model: Kimi K3.1" as it writes a procedural 3D scene in Three.js. Moonshot has not announced the model, a release date, pricing or weights.
MiniMax publishes the MiniMax M3.1 Flash weights
MiniMax says the weights of MiniMax M3.1 Flash are now public on Hugging Face and ModelScope. When it launched the model on October 19, MiniMax said the weights would follow later.GitHub Copilot in Visual Studio September update: PR exploration, BYOM changes, NuGet fixes
GitHub's September roundup for Copilot in Visual Studio 2026 lists three changes. It applies to all Copilot plans, including Free, Student, Pro, Pro+, Max, Business and Enterprise.Artificial Analysis releases Coding Agent Index v2.0; Claude Code with Opus 5.5 leads at 61.7
Artificial Analysis launched v2.0 of its Coding Agent Index, which scores models together with their agent harnesses on 290 tasks across software engineering, terminal use and repository Q&A, weighted equally. The scores are Artificial Analysis's own, not independently verified.
Grok Bot 0.75.2 adds cards in desktop chats and video generation
Grok Bot 0.75.2 is out, according to a post on X from a member of the Grok Bot team that links to the release's changelog on x.ai. The post lists three changes:Perplexity releases Decider for Mac, which runs its 4B decision model locally
Perplexity says you can now run pplx-decider-flash-v1.2 on your Mac without an API key. Decider for Mac is an app that runs the 4B decision model on the device: you ask a question about text or an image and get a probability back, all on your machine.
Copilot usage metrics API now names custom skills, agents and MCP servers
GitHub says Copilot usage metrics now show the configured names of the skills, custom agents, Model Context Protocol (MCP) servers, slash commands and plugins that an organization creates, instead of grouping them under other or custom. It builds on the agentic CLI customization metrics released in September.Claude Code 2.1.309 fixes sandbox and permission-check bypasses and changes how `sed -i` and `rm` are handled
Claude Code v2.1.309, released October 22, 2026, per its release notes:Grok Bot can now use Calendar, Reminders and Contacts on your phone
Grok Bot, xAI's assistant, can now use apps on a user's phone, including Calendar, Reminders and Contacts, according to the announcement on X. The company says it is rolling out now on iOS and Android. The post includes an image; it had 5,209 likes and 1,367 replies by October 22.