News
ResearchOpenAI pens AI safety agreement with four firmsReleasesOVHcloud unveils Olala mini, its first in-house LLMChinaA 9B model beats a 1T model on FinQA, trained with Crusoe Managed RLReleasesThe internal metaphysics of an LLMReleasesMapping Language Model Internal Representations to Physical Laws and Scientific ReasoningResearchCan China keep rogue AI in check? Meet the start-up testing the top modelsFrontierMicrosoft aims to challenge OpenAI, Anthropic with cheaper, specialized AI modelsOpen modelsMicrosoft rolls out new first-party models and updates to agentic AI continuous improvementFrontierOpenAI鈥檚 New Hacking Model Found Dozens Of Holes In Apple TechFrontierMicrosoft unveils AI models for coding, cybersecurity as it takes on Anthropic, OpenAIResearchOpenAI pens AI safety agreement with four firmsReleasesOVHcloud unveils Olala mini, its first in-house LLMChinaA 9B model beats a 1T model on FinQA, trained with Crusoe Managed RLReleasesThe internal metaphysics of an LLMReleasesMapping Language Model Internal Representations to Physical Laws and Scientific ReasoningResearchCan China keep rogue AI in check? Meet the start-up testing the top modelsFrontierMicrosoft aims to challenge OpenAI, Anthropic with cheaper, specialized AI modelsOpen modelsMicrosoft rolls out new first-party models and updates to agentic AI continuous improvementFrontierOpenAI鈥檚 New Hacking Model Found Dozens Of Holes In Apple TechFrontierMicrosoft unveils AI models for coding, cybersecurity as it takes on Anthropic, OpenAI
EveryLLM

Reasoning models

OpenAI o1 and o3, DeepSeek-R1 and the models that think step by step before they answer. 100 models on EveryLLM so far.

Reasoning models are language models trained to work through a problem in a long chain of intermediate steps before they give a final answer. The extra thinking tokens cost time and compute each time the model is used, but they lift results on mathematics, science and coding. OpenAI's o1, previewed in September 2024, made the approach mainstream, and DeepSeek-R1 in January 2025 showed that it could be reproduced with open weights.

Where the idea came from

Researchers at Google showed in a January 2022 paper that simply prompting a large model to write out its steps, which they called chain-of-thought prompting, improved its answers on arithmetic and logic problems. Reasoning models go further. They are trained with reinforcement learning, rewarded when a long chain of thought ends in a correct, checkable answer, so the habit of planning, checking and backtracking is built into the model itself.

The first wave

OpenAI released o1-preview on 12 September 2024 and reported that it solved 83% of problems on a qualifying exam for the International Mathematics Olympiad, against 13% for GPT-4o. DeepSeek-R1 followed on 20 January 2025 under the MIT licence, with a paper describing how reasoning emerged from large-scale reinforcement learning. Anthropic's Claude 3.7 Sonnet (February 2025) and later models let users choose how long the model thinks.

Costs and limits

A reasoning model can write thousands of hidden tokens for one question, so answers are slower and dearer to produce. Labs now train models that decide for themselves how much to think: GPT-5, released in August 2025, routes each request between a fast model and a deeper reasoning model. Researchers also debate how faithfully the visible chain of thought reflects what the network actually computes.

Famous reasoning models

Further reading

MiMo-V2-Flash 馃嚚馃嚦 ChinaMiMo-V2-Flash High-speed MoE language model for reasoning and agentsGeneral 路 Open weights fnNemotron Nano 2 馃嚭馃嚫 United StatesNemotron Nano 2 A 9B hybrid Mamba-Transformer model for reasoning and chat, from NVIDIA.General 路 Open weights 路 2025 fnNemotron 3 Super 馃嚭馃嚫 United StatesNemotron 3 Super NVIDIA's 120B hybrid MoE language model with 12B active parameters and 1M-token contextGeneral 路 Conditional licence 路 2026 fnEXAONE 4.0 馃嚢馃嚪 South KoreaEXAONE 4.0 Unified language model with reasoning and non-reasoning modesGeneral 路 Conditional licence 路 2025 fnSolar Open 2 馃嚢馃嚪 South KoreaSolar Open 2 250B parameter open-weight model for agentic tasksGeneral 路 Open weights 路 2026 Qwen3.8-2.4T-A95B 馃嚚馃嚦 ChinaQwen3.8-2.4T-A95B Open-weight reasoning model with 2.4 trillion parametersReasoning 路 Open weights 路 2026 Falcon-H1R-7B 馃嚘馃嚜 United Arab EmiratesFalcon-H1R-7B A 7-billion parameter reasoning model with hybrid architectureReasoning 路 Open weights 路 2026 Falcon-H1 馃嚘馃嚜 United Arab EmiratesFalcon-H1 Hybrid transformer and state-space model series by TIIGeneral 路 Open weights 路 2025 fnMiniMax-M1 馃嚚馃嚦 ChinaMiniMax-M1 Open-weight reasoning model with hybrid attention and long contextReasoning 路 Open weights 路 2025 Hunyuan-TurboS 馃嚚馃嚦 ChinaHunyuan-TurboS Hybrid Mamba-Transformer MoE model by TencentGeneral 路 API only 路 2025 imgvidNemotron 3.5 Super VL 馃嚭馃嚫 United StatesNemotron 3.5 Super VL NVIDIA's open-weight vision-language model with 120B parameters and 1M-token contextMultimodal 路 Conditional licence 路 2026 fnNemotron 3 Nano 馃嚭馃嚫 United StatesNemotron 3 Nano NVIDIA's 30B hybrid Mamba-Transformer MoE model for reasoning and agentsGeneral 路 Conditional licence 路 2025

The weekly digest

Every Monday: the model of the day, new pages on EveryLLM, and the conferences and anniversaries of the week ahead.