X CognitionCognition launched private coding benchmark FrontierCode 2.0 with 132 real open-source tasks. Claude Opus 5.5 leads at 46%; DeepSeek V4.1 Flash tops open weights at 28%.
View originalX CognitionArtificial Analysis launches Intelligence Index v5, with 60% of evaluations on private datasets. Claude Opus 5.5 tops at 58, GPT-6 Astra at 57; open-weights gap widens to 12 points.
View original
X AnthropicAnthropic signs OneGov agreement with GSA to bring Claude to federal, state, local, tribal and territorial governments at a 40% discount.
View originalX AnthropicAnthropic scientists partnered with six drug developers and launched internal drug discovery efforts, sharing early results on Claude's role across drug development stages.
View original
X GeminiGoogle launches SynthID detection in Gemini, letting users ask 'Is this AI generated?' to identify AI-made images, videos, and audio.
View original
X CursorCursor's design mode adds Annotate, enabling visual prompts, element editing, on-page drawing, and responsive design review.
View original
X Kling AIMiora launches Halloween Pop-Up Challenge with up to 100,000 Credits in prizes, recommending Kling 4.0 for AI video creation; ends Nov 3.
View original
HuggingFace BlogLiquid AI LFM3: new 2B family across text/vision/audio/omni/decision; edge-ready, tune on one GPU.
View originalHuggingFace BlogAi2 releases study showing LLM novelty judges are unreliable: small prompt changes swing accuracy by 50 points; retrieval and reasoning do not improve.
View original
X OpenAI DevelopersOpenAI launched Decisions API in public beta for all developers, claiming decisions up to 10x faster than GPT-6 Luna via Responses API, enabling live command scoring.
View originalGemini BlogGoogle showcases eight applications of Gemini 4 Argon, its most capable model, from deep research to interactive design.
View originalOpenAI NewsRetell: 100M calls/mo via OpenAI; GPT-6 Astra tops voice benchmark; Codex - $1.6M rev/employee.
View original
Cloudflare BlogCloudflare launches a PHP CTF challenge integrated with its Bug Bounty program, inviting security researchers to bypass WAF defenses and help make the Internet safer.
View originalOpenAI NewsOpenAI and Oracle expand partnership to bring business apps and AI agents into ChatGPT.
View originalX CursorDeepSeek V4.1 Flash is live in Cursor, setting the CursorBench cost frontier at $0.60–$0.98/task.
View originalCloudflare BlogA configuration mismatch in GLM 5.3 on Workers AI triggered an SGLang bug exposing KV cache pages across requests; remediation has been implemented.
View originalOpenAI NewsGreptile uses GPT-6 Astra to extend bug detection beyond code review to user workflows, with Daybreak for security and Codex to reduce rework.
View originalX OpenAI DevelopersGPT-6.1 Sol, with CodeRabbitAI, helps developers find security risks in complex codebases.
View original
X Sam AltmanOpenAI announces three new discoveries in harmonic analysis, complexity theory, and quantum field theory.
View originalX OpenAI DevelopersIn a side-by-side test, GPT-6.1 Terra filled a form in 53.5s vs Sol's 3:21.7, a 2:28.2 win.
View original