LLM

A LocalLLaMA discussion of SWE-rebench January runs reports close top-tier results, with Claude Code leading pass@1 and pass@5 while open models narrow the gap.

Google announced a major Gemini 3 Deep Think upgrade with stronger reasoning benchmarks and early API access for researchers and enterprises.

OpenAI announced on February 9 that it's testing ads in ChatGPT for Free and Go tier users in the US. Plus, Pro, Business, and Enterprise tiers remain ad-free.

Anthropic released Claude Opus 4.6, achieving industry-leading performance in coding, long-context retrieval, and knowledge work.

OpenAI disbanded its Mission Alignment team, which communicated the company's mission to the public and employees. The team leader was reassigned as 'Chief Futurist' amid renewed AI safety concerns.

Anthropic raised $30B at a $380B valuation and now leads the enterprise LLM market with 32% share, surpassing OpenAI's 25%.

Major U.S. grocery chain Albertsons joined OpenAI's ChatGPT advertising pilot. The test explores conversational AI ad formats for retail, signaling growing industry interest in AI-native advertising.

Microsoft AI Safety team discovered GRP-Obliteration, an attack that disables safety alignment across 15 major LLMs with a single prompt. GPT-OSS-20B's attack success rate jumped from 13% to 93%.

Meta has unveiled Llama 4 Scout and Maverick, the first open-weight natively multimodal models. With industry-leading 10 million token context and MoE architecture, they outperform GPT-4o and Gemini 2.0 Flash.

DeepSeek is set to launch its next-generation coding-focused AI model V4 in mid-February, featuring 1M+ token context windows and consumer GPU support for unprecedented developer accessibility.

Z.ai unveiled GLM-5, a 744B parameter (40B active) model pre-trained on 28.5T tokens. Designed for complex systems engineering and long-horizon agentic tasks, it leads open-source models in multiple benchmarks.

OpenAI launches GPT-5.3-Codex, the first model to debug its own training and manage deployment. Released with tight security controls due to cybersecurity concerns.