Fable 5.2 preserved thinking now binds every API account
Anthropic now enforces preserved thinking on every Fable 5.2 API account, so editing earlier turns can return a 400 unless the harness uses a supported path.
Anthropic now enforces preserved thinking on every Fable 5.2 API account, so editing earlier turns can return a 400 unless the harness uses a supported path.
OpenAI’s Control Room tracks research inference jobs, including agentic sampling, and can halt further model output within 35 seconds of a stop.
Anthropic ships Fable 5.2 at $8/$40, up to 30% faster than 5.1, with a 2.5x fast mode and Bedrock availability for long-horizon agents.
OpenAI lists ChatGPT Enterprise in AWS Marketplace so buyers can use existing AWS purchasing and apply eligible spend to commitments.
Apiiro counted 2,539 secrets in public GitHub repositories after Claude Code saved approved commands, including credentials, into project settings files.
X Post Facto · 1–7 Nov 2026
Open WebUI 0.13.1 adds the Tether desktop companion and lets Claude Code, Codex, and Pi appear as models in Open Terminal.
Agent Framework Python 1.25.0 drops Python 3.10, caps MCP discovery, and binds each live workflow session to one authorization scope.
ADK Python 2.14.0 graduates environments, A2A, and eval from experimental, and changes how Enum tool arguments arrive.
Anthropic’s research post explains Claude’s session notes, salience, and an offline dreaming pass that checks memory against transcripts.
Meta is rolling Active Memory and Active Focus into Muse and Meta AI, citing fewer forgotten goals and better recovery when plans change.
Ant Ling released Ling-3.2-flash, a 560B/25B multimodal MoE with a 1M window, aimed at long-horizon visual and agent tasks.
GPT-6 Cyber is a Critical-rated security model for approved Daybreak Red customers, first available through the API and Codex.
langgraph-sdk 0.4.14 adds strict_resume on run create, stream, and wait, plus tools and lifecycle stream modes.
OpenAI and Ligo’s BioBasicsBench scores agents on 100 routine wet-lab tasks. GPT-6 Astra leads at 66.7% partial credit.
OpenAI is partnering with Hacktron to test its systems, especially third-party service connections, and is expanding the firm’s Daybreak Red model access.
Claude Code 2.1.319 denies or asks about auto-mode calls a gateway cannot check, and backgrounds MCP tasks in the terminal.
Linear’s account posted a Grok Bot product-ops setup that uses its cloud coding agent, with $25 in AI credits for 30 days.
Cat Wu, who led product for Claude Code, is joining OpenAI to lead product management for Core Products, OpenAI confirmed to Business Insider.
OpenAI’s November 4, 2026 write-up says one upcoming cyber-model run drafted patches for 948 unique Linux local privilege-escalation bugs, at about $1,279 each, after human review—with only 61 patched upstream when the 10-day window closed.
Adobe says CX Enterprise Coworker now runs in Claude, Gemini, Copilot, and Databricks as well as ChatGPT, with new MCP connectors to ads and data tools.
Images 2.6 is live in ChatGPT and Codex on every tier, with GIFs plus Flare and Sunburst models in the API.
Approved customers can use Daybreak Red's GPT-5.6 Cyber in AWS GovCloud (US-East) on Bedrock; AWS had already posted it Oct. 30.
Anthropic expanded Claude Marketplace on Nov. 4 with partner agents and products, and eligible enterprises can apply committed spend.
LangGraph 1.2.20 gives each interrupt() in a task its own ID and adds a strict resume mode that refuses stale interrupt answers.