GameDuo reduced observability costs by more than half and extended log retention to 180 days by migrating to ClickStack.
The 13B member of the family uses 91% of the larger parent's compute while maintaining a base model status.
The engine allows AI agents to write standard React and TypeScript code that compiles to GPU-accelerated video, motion graphics, and interactive 3D games, with a 90-frame render taking about one second on Apple Silicon.
TVM and Together Compiler demonstrate that stable IRs and verification tools are necessary to constrain search costs and ensure correctness when AI agents optimize inference kernels.
The 1.0 release adds a modernization workflow, expanded guide database, and public eval dashboard, with Firefox and Edge joining the cross-browser compatibility initiative.
A test on a real bug showed Opus 5.5 high failed to find a fix after 3 hours, while xhigh resolved it in minutes.
NVIDIA releases GR00T 2, a foundation world-action model that predicts future video and robot actions, enabling zero-shot transfer from human demos and real-time control of Unitree G1 humanoids via the GR00T 2 Turbo variant.
The system integrates with inference engines to deny requests for stopped jobs, ensuring a fail-closed state if the service goes down.
The framework compresses Nemotron 3.5 Super from 120B to 74B parameters while retaining 95% of its mean benchmark score.
The paper distinguishes standard test-time scaling from compute extrapolation, where performance continues to improve at test-time compute levels far exceeding the maximum used during training.
The GhostShares investigation found that Grok's default 'Anyone with the link' setting and DNS-discoverable hostnames exposed 35,578 apps, affecting 30 external organizations and SpaceXSI/X staff.
Analysis of Qwen3.6-35B-A3B shows that ablating 92.2% of weak neuron activations causes minimal performance loss, proving neurons are tractable units of analysis.
The method learns query-dependent weights to compose cached state encodings, improving performance on document QA, needle-in-a-haystack retrieval, and tool use.
The study demonstrates that gating mechanisms in SSMs cause models to learn in-weights memorization solutions, delaying or preventing convergence to correct in-context learning even without architectural memory limits.
Study shows judge quality varies by task and criteria, with paired comparisons outperforming independent scoring and top models exhibiting exaggerated sensitivity to subtle details.
The startup combines GPT-6.1 Sol and GPT-5.6 with enterprise context to autonomously investigate and fix security risks, achieving 5x triage throughput for IT services clients.
The new models support custom dimensions and aspect ratios, with the Highspeed variant offering faster generation using the same API key.
Football Manager 27 Touch is added to Apple Arcade, and four new titles are released on December 3.
The agent handles one provider deal at a time on Claude Opus, supported by Sonnet and Haiku agents for rate review and message checking.
The new features allow teams to build spreadsheets, whiteboards, and dashboards directly in Claude, with interactive connectors enabling work inside apps like Adobe and Atlassian without switching tabs.
The feature rolls out on web and desktop for all plans, starting with tools like Asana, Canva, and Dropbox via the MCP Apps extension.
Organizations can now link their Hugging Face account to Google Cloud to consolidate billing and avoid new vendor onboarding.
The model features a 1M-token context window for complex software engineering and cybersecurity tasks, with a 50% price reduction ending November 30, 2026.
The new interface supports choice, score, and noul primitives with a unified API across transformers and llama.cpp.
AI Briefing