Follow the AI agents and models you build with

Gemini 4 Carbon generally available, introductory pricing runs through November 30

Gemini 4 Carbon, described as the most capable reasoning model for sustained agentic coding and cybersecurity defense, is now generally available in the Gemini API. Introductory pricing is $2.00 per 1M input tokens and $10.00 per 1M output tokens through November 30, 2026.

Claude Fable 5.2 launches with a 1M token context window and 128k max output

Claude Fable 5.2 is built for demanding reasoning and long-horizon agentic work. It has a 1M token context window, 128k max output tokens, and always-on adaptive thinking.

Gemini Decision launches with the Decisions API for zero-output-token decisions

Google released Gemini Decision and the Decisions API for fast, calibrated binary, categorical, and rubric decisions across text, images, audio, video, and PDFs with zero output tokens.

Cursor Projects goes multiplayer in beta for all users

Projects now lets teammates join a shared Project with a shared context system and agent-to-agent communication. It is rolling out in beta to all Cursor users starting today.

Gemini 4 Carbon debuts 1st of 420 on Arena text

In its first result on Arena text, Gemini 4 Carbon at high effort scored 1526.6, 1st of 420.

Claude Fable 5.2 and Mythos 5.2 return 400 errors when the prefix changes

Both models enforce the prefix check for every account. Requests that edit the system prompt, tools, or an earlier message, then send back a thinking block from a later turn, return a 400 error.

Suggested next step:

Check whether your code edits the prefix before you switch models.

Claude Opus 5.5 users report nerfed thinking patterns since November 3

A Hacker News thread asks whether Opus 5.5 was nerfed because its thinking patterns changed on November 3. The thread has 320 points and 217 comments.

Claude Fable 5.2 costs $8 per million input tokens, down from $10

Claude Fable 5.2 costs $8 USD per MTok of input and $40 USD per MTok of output. Claude Fable 5.1 cost $10 USD per MTok of input and $50 USD per MTok of output.

Claude Fable 5.2 becomes available in GitHub Copilot

GitHub Copilot adds Claude Fable 5.2 as a model option.

Claude Fable 5.2 and Mythos 5.2 launch with outputs up to 30% faster

Claude Fable 5.2 and Claude Mythos 5.2 are upgrades that show improvements in coding, reasoning, and vision. Testers found they generate outputs up to 30% faster than Claude Fable 5.1 and Claude Mythos 5.1.

Meta releases Muse Ember, its smallest model, at $0.10 per million input tokens

Muse Ember is available today on Muse Code and Meta Model API. It costs $0.10 per million input tokens, $0.004 per million cached input tokens and $0.50 per million output tokens, with a 1M-token context window and up to 512K output tokens.

Gemini 4 Carbon debuts 2nd of 57 on Arena agent

In its first result on Arena agent, Gemini 4 Carbon at high effort scored 0.2, 2nd of 57.

ChatGPT Pro plans have no 5-hour limits and give 10X, 20X and 50X Plus usage

OpenAI clarified that all Pro plans have no 5-hour limits. Pro 100 gives 10X, Pro 200 gives 20X and Pro 500 gives 50X the Plus allowance, with access to ultrafast on Pro 500.

Gemini 4 Carbon debuts 3rd of 153 on Arena WebDev

In its first result on Arena WebDev, Gemini 4 Carbon at high effort scored 1768.1, 3rd of 153.

GPT Image 2 is deprecated and shuts down in the API on February 10, 2027

The gpt-image-2 model and its 2026-04-21 snapshot will be removed from the API on February 10, 2027. OpenAI recommends gpt-image-2.6-sunburst or gpt-image-2.6-flare as replacements.

Takes effect Feb 10, 2027 · in 90 days

Suggested next step:

Move image generation calls to gpt-image-2.6-sunburst or gpt-image-2.6-flare before February 10, 2027.

Claude Code auto mode now asks or denies unchecked calls without telemetry

With telemetry or feature flags off, including behind LLM gateways, calls the server can't check are asked about or denied. There is no fallback to the built-in classifier.

Suggested next step:

Set CLAUDE_CODE_AUTO_MODE_SERVER=0 to keep using the built-in classifier.

OpenAI releases GPT-6-Cyber in the API for vulnerability research

GPT-6-Cyber is available in the API to investigate vulnerabilities, reproduce them in an authorized test environment, and verify fixes. Daybreak setup is required before the first request.

Claude adds Personal Finance beta for US Pro, Max and Ultra plans

Personal Finance is a beta that lets users link bank, credit card, loan, and investment accounts to Claude to see balances and spending on one page and ask questions about their money. It is available on Pro, Max, and Ultra plans in the US only.

Cursor adds live app building with database, auth and hosting on November 5

Cloud agents can now build a web app, set up a Postgres database, configure auth behind Cursor SSO, and deploy it to a shareable link. Publishing runs security checks first, and team admins can see, review and unpublish every app.

OpenAI releases GPT-Image-2.6 Sunburst and Flare with SVG and GIF generation

GPT-Image-2.6 Sunburst and Flare are released for image generation and editing, with adjustable complexity, SVG generation, and animated GIF generation. Markup-guided editing is added for supported GPT Image 2.5 and 2.6 models.