AI Daily · Oct 29, 2026 · Thu
ICREAT AI NEWS
Google DeepMind Releases Gemini 4 Carbon Frontier Model
Google DeepMind releases Gemini 4 Carbon, claiming it achieves state of the art on cybersecurity and long-horizon tasks, ranking first with 59 points on the Artificial Analysis Intelligence Index v5.0.
- Gemini 4 Carbon Tops the Artificial Analysis Intelligence Index v5Artificial Analysis Articles
Model Releases/Updates
NVIDIA Releases Nemotron 3.5 Transcribe and Granary v2 Data Pipeline
NVIDIA releases the open-weight model Nemotron 3.5 Transcribe on Hugging Face and ModelScope, combining transcription, speaker diarization, translation, and audio question answering into a single checkpoint, supporting transcription in 40 languages, translation between 31 languages and English, speaker labeling in 13 languages, and processing recordings up to 1 hour long with up to 8 speakers labeled.
Product Releases/Updates
Meta Releases Muse Realtime Avatar, Turning Any Image into a Real-Time Conversational Digital Human
Meta releases Muse Realtime Avatar, driving the voice token stream of Muse Realtime Voice into a real-time digital human with facial expressions, gestures, and full-body motion, with any image usable as a character reference.
GitHub Copilot Fusion Schedules Anthropic and OpenAI Models to Balance Performance and Cost
GitHub releases Copilot Fusion in GitHub Copilot, performing runtime scheduling across multiple Anthropic and OpenAI models to optimize performance, cost, and latency per task.
Industry news
The DecoderSecurity Startup Cathedral Used AI Agents to Breach Anthropic's Internal Systems
Security startup Cathedral claims it used AI agents to enter Anthropic's internal systems through third-party business software, reaching meeting notes, payroll data, recruiting information, and Slack messages in under twelve hours.
Optimus Labs Retrospectives Incident in Which AI Agents Achieved Code Execution via RubyGems and RubyDoc.info
Optimus Labs retrospectives an incident in which AI agents submitted over 2,000 gems to RubyGems in about 48 hours, exploiting Ruby code pointed to by .yardopts to execute on the build host when RubyDoc.info built documentation, then packaging scraped public site content into new gems and sending them back.
Paper Research
Artificial Analysis Releases Intelligence Index v5 and Progress Index v1
Artificial Analysis releases Intelligence Index v5, adding two new benchmarks, FrontierCode 2.0 and Terminal-Bench-Science, upgrading HLE-Diamond, AA-Omniscience v1.1, and CritPt v2, with 60% of evaluations based on private test sets or held-out answers.
Replit BlogFirst partyReplit Releases ViBench 2.0 Long-Horizon Software Building Benchmark
Replit releases ViBench 2.0 to evaluate agents' long-horizon software development ability to build a complete product feature by feature starting from an empty project, comprising 17 products, 241 feature requests, and 544 test plans.
Tips and Opinions
Pydantic AI ArticlesFirst partyPydantic AI Tests Code Mode: Long Tasks Cost 58% Less, Small Tasks Actually More Expensive
The Pydantic AI team compared code mode with regular tool calling across 5 small applications, 8 models, and 5,276 scored runs, finding that on long tasks cost dropped 58%, time dropped 51%, and success rate rose 16 percentage points.