33 stories tagged reasoning-models, newest first.
103 points by armcat | 19 comments
BottleCap AI blog post announcing ThinkingCap-Gemma-4-26B-A4B, a fine-tune cutting thinking tokens by 39% on average with near-unchanged benchmark accuracy.
232 points by nicowaltz | 89 comments
GitHub repository from PostHog for Jeeves, a project showing that reasoning improves Jev-like decision models.
479 points by leumon | 316 comments
Google blog post announcing Gemini 3.8 Live and 3.8 Live Extended Thinking, live voice dialogue models with improved intelligence and parallel reasoning.
732 points by bilsbie | 343 comments
Simon Willison blog post reviewing Alibaba's Apache 2 licensed Qwen 3.8 27B vision LLM run locally, noting strong results but excessive default reasoning.
761 points by tosh | 454 comments
ARC Prize verified results page for DeepSeek V4 Flash 0731, reporting its ARC-AGI-1 and ARC-AGI-2 scores and cost per task on a leaderboard chart.
330 points by fittingopposite | 121 comments
Inception Labs blog post by CEO Stefano Ermon announcing Mercury 2, a fast reasoning LLM built on diffusion rather than autoregressive generation.
1030 points by tosh | 681 comments
Google DeepMind post announcing an upgraded Gemini 3 Deep Think reasoning mode for science and engineering, featuring Duke's Wang Lab designing semiconductor materials.
471 points by vinhnx | 410 comments
Qwen team blog post announcing Qwen3-Max-Thinking, a reasoning-focused version of their flagship large language model.
339 points by janpio | 107 comments
Google blog post announcing Gemini 3 for developers, covering its new reasoning and agentic capabilities.
843 points by nekofneko | 381 comments
Moonshot AI's announcement of Kimi K2 Thinking, an open-source reasoning agent model that can run 200–300 sequential tool calls and scores highly on HLE and BrowseComp.
487 points by Terretta | 463 comments
xAI announcement of grok-code-fast-1, a fast and economical reasoning model designed for agentic coding.
733 points by wertyk | 253 comments
DeepSeek API docs news post announcing the DeepSeek-V3.1 release, featuring hybrid think/non-think inference, faster reasoning, and stronger agent skills.
2027 points by lackoftactics | 804 comments
OpenAI's landing page for its open-weight reasoning models (gpt-oss), with links to download on Hugging Face, view on GitHub, and try the models.
325 points by danboarder | 70 comments
GitHub repository for MiniMax-M1, an open-weight, large-scale hybrid-attention reasoning language model released under Apache-2.0 with docs and tech report.
893 points by meetpateltech | 395 comments
Mistral AI's announcement page for Magistral, its first reasoning language model.
361 points by error404x | 182 comments
Hugging Face model page for DeepSeek-R1-0528, a 685B-parameter text generation model from DeepSeek, shown without a model card yet.
386 points by codelion | 62 comments
Show HN post introducing AutoThink, a technique that adaptively allocates thinking tokens by query complexity and uses steering vectors to improve local LLM reasoning.
465 points by thm | 178 comments
GitHub repository for Xiaomi's MiMo-7B, a language model aimed at reasoning, with a technical report covering pretraining through posttraining.
808 points by synthwave | 369 comments
Blog post from the Qwen team announcing the Qwen3 family of large language models with hybrid thinking and non-thinking modes.
369 points by meetpateltech | 246 comments
Anthropic research post on whether reasoning models' Chain-of-Thought faithfully reflects their actual reasoning, linking to the accompanying paper.
910 points by meetpateltech | 450 comments
Google blog post announcing Gemini 2.5 Pro Experimental, a thinking model that leads common benchmarks with strong reasoning and coding capabilities.
445 points by nwjsmith | 147 comments
Qwen Team blog post announcing QwQ-32B, a 32B-parameter reasoning model trained with scaled reinforcement learning.
399 points by 9woc | 62 comments
UC Berkeley Sky Computing Lab project page for Sky-T1, a reasoning model trained for under $450 that rivals OpenAI's o1-preview.
442 points by sebg | 163 comments
Sebastian Raschka's Ahead of AI newsletter article explaining the four main approaches to building and refining reasoning LLMs.
796 points by tkellogg | 365 comments
Blog post by Tim Kellogg discussing S1, a reasoning model reportedly trained for about $6 that competes with DeepSeek R1.
915 points by johnneville | 861 comments
OpenAI's announcement of o3-mini, a cost-efficient reasoning model available in ChatGPT and the API, strong in science, math, and coding.
703 points by meetpateltech | 257 comments
ARC Prize blog post by Mike Knoop analyzing DeepSeek's R1-Zero and R1 models, arguing R1-Zero is more important, with ARC-AGI benchmark results.
354 points by jonbaer | 186 comments
Hugging Face blog post announcing Open-R1, a project to build a fully open reproduction of the DeepSeek-R1 reasoning model.
472 points by amrrs | 107 comments
Jay Alammar's illustrated explainer newsletter post on how DeepSeek-R1 was trained, presented as a recipe for reasoning LLMs.
1582 points by meetpateltech | 585 comments
GitHub repository for DeepSeek-R1, DeepSeek's open reasoning language model, with the paper PDF, figures, README, and MIT license.
871 points by simonw | 530 comments
Simon Willison's blog post reviewing the major developments, themes and pivotal moments in large language models during 2024.
1578 points by maurycy | 1581 comments
ARC Prize blog post by François Chollet reporting OpenAI's o3 scoring 75.7% (and 87.5% at high compute) on the ARC-AGI semi-private evaluation set.
358 points by amrrs | 260 comments
Qwen Team blog post announcing QwQ-32B-Preview, an experimental open reasoning-focused language model from Alibaba similar to OpenAI's o1.