Mistral Small 4
Mistral AI's unified MoE model - 119B total parameters, 6B active per token, 128 experts, 256K context, configurable reasoning, Apache 2.0 license.
Articles Tagged "Reasoning"
Mistral Small 4: 128 Experts, 6B Active, Apache 2.0
Mistral AI releases Small 4 - a 119B MoE with only 6B active parameters, 256K context, configurable reasoning, and Apache 2.0 license. Plus a new NVIDIA partnership to co-develop frontier open models.
Balanced Thinking, Broken Judges, Opaque Reasoning
Three new papers expose cracks in how AI models think, how benchmarks evaluate multimodal reasoning, and why LLM judges reliably mislead.
Reasoning Traps, LLM Chaos, and Steering Curves
Three papers this week: why better reasoning creates safety risks, why multi-agent systems behave chaotically even at zero temperature, and why straight-line activation steering is broken.
NVIDIA Nemotron 3 Super 120B-A12B
NVIDIA Nemotron 3 Super is a 120B-parameter open model with 12B active at inference, combining Mamba-2, LatentMoE, and Multi-Token Prediction for agentic workloads with a 1M token context window.
VLMs Fail Physics Tests, RL Quits Bad Paths, Agents Lie
Three new papers expose systematic VLM failures on basic physics, introduce RL that learns to abandon bad reasoning paths, and reveal that AI agents deceive primarily through misdirection rather than fabrication.
Grok 4
Grok 4 is xAI's frontier reasoning model, the first to break 50% on Humanity's Last Exam, with a 256K context window, $3/M input pricing, and a Heavy multi-agent variant built on 200,000 GPUs.
Reasoning Models Can't Hide Their Thinking - OpenAI Study
OpenAI's CoT-Control benchmark shows frontier reasoning models score 0.1-15.4% at steering their own chain of thought - a result the company frames as good news for AI oversight.
CoT Control, Hidden Beliefs, and Dynamic Agent Benchmarks
New research shows reasoning models can't suppress their chain-of-thought, that they commit to answers internally long before their CoT reveals it, and that static benchmarks are inadequate for measuring real-world agent adaptability.
Claude Opus Reasoning Distilled Into Open 27B Model
A community fine-tune distills Claude Opus 4.6 chain-of-thought reasoning into Qwen3.5-27B via LoRA, racking up 4,000+ downloads in days. No benchmarks yet - but the approach raises familiar questions.
Qwen3.5-27B Claude Opus Distilled
Community fine-tune that distills Claude Opus 4.6 reasoning into Qwen3.5-27B via LoRA. 28B parameters, Apache 2.0, no published benchmarks.
Qwen3.5-27B Distilled vs Base: What You Gain
Comparing the Claude Opus reasoning-distilled Qwen3.5-27B against the base model - what chain-of-thought distillation adds and what it costs in context, multimodal, and reliability.
