Gemini Robotics ER 2 introduces native video understanding and multi-robot orchestration, bypassing traditional text-based planning bottlenecks.
Anthropic's models successfully exploited real-world software vulnerabilities during security testing, proving autonomous offensive cyber capabilities are active.
Dialog-RSN-1 processes raw audio directly, eliminating ASR latency to unify turn-taking and response generation in one model.
KotlinLLM generates and hot-reloads Kotlin source code at runtime, replacing slow live API calls with local compiled macros.
Distilling DeepSeek into GPT-OSS demonstrates that safety alignment and censorship do not transfer during standard distillation workflows.
Google used LLMs to patch more Chrome vulnerabilities in one month than the previous two years combined.
The LLM CLI tool updates its default model to GPT-5.6 Luna and restructures schema logging for complex agent interactions.
Science One introduces a verifiable autonomous research framework that enforces a strict chain-of-evidence to prevent hallucinated scientific discoveries.
Nous Research integrates Hermes Agent with Nostr, establishing a decentralized, censorship-resistant communication protocol for human-agent collaboration.
This local merge queue prevents race conditions and conflicts when running multiple Claude Code agents on the same codebase.
Analysis of DeepSeek V4 Flash reveals the current frontier of open-weights cost-to-performance ratios for high-throughput production pipelines.
This Go SDK simplifies production deployments of streaming, tool-calling AI backends with native React frontend bindings.