Reasoning Models
Reasoning models are a new class of AI that "thinks" before answering: spending extra compute at inference time, building chains of thought and solving problems ordinary LLMs cannot — math, code, science. They kicked off the test-time-compute race. Here is all our coverage of reasoning models: releases, benchmarks (GPQA, ARC-AGI), limits and cost.
LISA: Sparse Attention Speeds Up Long LLM Reasoning by 50% Without Retraining
The LISA method replaces the standard attention mechanism in language models and speeds up long reasoning by 50% at a 16K-token context, with no training from…
Tencent Hunyuan's Multimodal AI Chief Hu Han Departs as Team Pivots to World Models
Hu Han, who led multimodal understanding at Tencent Hunyuan, is leaving to found a startup; his group is being redirected to world models research.
China's Open AI Models vs. OpenAI and Anthropic: What DeepSeek and Qwen Are Offering
China's DeepSeek, Qwen, Kimi and GLM labs are pitching open models as a replacement for closed OpenAI and Anthropic flagships as access to Western models gets…
How Context Engineering Makes Weak Local LLM a Reliable Agent
Nvidia rival Etched raises $800M from Jane Street and TSMC partner venture fund
Vercel Released AI SDK 7 with New Tools for Managing AI Agents
LangChain: claude-3.5-sonnet, gpt-4o, o1, and o3-mini Tested as ReAct Agents
NVIDIA Developer Blog Explains AI Agent Reinforcement Learning Techniques
NVIDIA Developer Blog published a breakdown of reinforcement learning techniques for AI agents—from classic RLHF to new feedback-based model training approaches.
OpenAI released flagship model GPT-5.6 in three versions — Luna, Terra and Sol
OpenAI released GPT-5.6 in three versions — Luna, Terra and Sol — and claimed that the senior model outperformed Claude Fable 5 by 13.1 points on the agent tasks…
NVIDIA Cosmos 3: models for reasoning and actions in robots' physical world
Nemotron 3 Ultra Reaches Claude Opus Level 8 Times Cheaper
DeepSeek V3.2 Achieves 67% Accuracy on the ARC-AGI Abstract Reasoning Benchmark
AWS published 6 approaches to optimizing MCP tools to save context
Meta released Muse Spark 1.1 — multimodal agent with 1 million token context
Meta Superintelligence Labs introduced Muse Spark 1.1 — an agentic model with a 1 million token context window, video and audio support, and an OpenAI-compatible…
Why Open AI's Growth Doesn't (Yet) Threaten Anthropic: A Lifecycle Analysis
TechCrunch explains the paradox: open models set records, but Anthropic continues to grow — because each player occupies its own phase of a single lifecycle.