Thinking & Research

OpenAI's flagship multimodal reasoning model with long-context tool use.

Frontier reasoning model tuned for math, science and coding workflows.

Anthropic's best coding and agentic model, strong at long autonomous tasks.

Top-tier reasoning model for research, analysis and complex writing.

Long-context multimodal model with native tool use and 1M+ token window.

xAI's flagship reasoning model with real-time X knowledge and tool use.

High-performance open MoE LLM rivaling closed frontier models on benchmarks.

Open reasoning model trained with RL, competitive with o1-class systems.

Open multilingual model family with hybrid thinking modes.

Small language model punching above its weight on reasoning benchmarks.

Predicts the structure and interactions of life's molecules.

Medical LLM achieving expert-level performance on USMLE-style questions.

Genome-scale foundation model spanning DNA, RNA and proteins.

Finance-domain LLM trained on decades of market and news data.

Improved GPT-4 series model with stronger coding and instruction following.

Compact reasoning model balancing cost and quality.

OpenAI's first open-weights reasoning models since GPT-2.

Hybrid reasoning model with extended thinking mode.

Multi-agent variant of Grok 4 for the hardest problems.

Open agentic foundation model from Zhipu's GLM family.

Trillion-parameter open MoE model with strong agentic skills.

Open reasoning model with 1M-token context.

Fast, low-cost frontier-tier LLM from 01.AI.

Baidu's flagship multimodal foundation model.

Frontier protein language model for biology design.

Open diffusion model for de novo protein design.

Open AlphaFold3-class biomolecular structure prediction model.

Best-in-class medium-range global weather forecasting AI.

Probabilistic AI weather forecasting beating ENS.

LLM tuned for therapeutic and drug development tasks.

Multimodal frontier model with open weights.

NVIDIA's open LLM family for synthetic data and reasoning.

End-to-end AI infrastructure stack combining PowerEdge servers, storage, networking and NVIDIA AI Enterprise.

Datacenter accelerator (CDNA 4) for training and inference of frontier models.

Managed AI training service on dedicated NVIDIA Hopper/Blackwell clusters.

Flagship datacenter GPU for trillion-parameter AI training and inference.

Datacenter AI accelerator targeting price/performance vs H100.

Custom AI accelerators powering Gemini training and Google Cloud AI.

Turnkey on-prem AI cloud built with NVIDIA, co-engineered for enterprises.

AI-ready storage stack co-engineered with NVIDIA for training pipelines.

Converged AI infrastructure with ONTAP storage and NVIDIA compute.

Liquid-cooled GPU SuperClusters for trillion-parameter LLM training.

Wafer-scale AI processor delivering record-breaking inference throughput.

Language Processing Unit delivering ultra-low-latency LLM inference.

Full-stack AI platform with Reconfigurable Dataflow Units (RDUs).

Open RISC-V based AI accelerator from Jim Keller's team.

Custom AWS chip purpose-built for training large language models.

Cost-optimized AWS chip for high-throughput LLM inference.

AI Platform deploying LLMs against classified and government datasets with audit and policy controls.

LLM-powered decision-making platform for defense and intelligence analysts.

AI for open-source intelligence and non-traditional collection.

Domain-specific large language models tuned for the Now Platform.

Generative analytics and natural-language insights inside Tableau.

In-warehouse ML and GenAI directly on BigQuery data via SQL.

Natural-language SQL and analytics assistant inside Snowflake.

Conversational analytics over governed lakehouse data.

Generative analytics service delivering trusted answers from unstructured data.

Analytics and AI platform with embedded LLM orchestration and copilots.

In-database analytics and GenAI orchestration on Teradata VantageCloud.

Anthropic's most intelligent model — state-of-the-art on coding, agents and computer use.

Frontier agentic coding model for long-horizon software engineering inside Codex.

Updated GPT-5 with warmer tone, adaptive reasoning and stronger instruction following.

Leads LMArena Text, WebDev and Vision — Google's flagship multimodal reasoning model.

Extended-thinking variant of Gemini 3 for hardest math, science and research problems.

Fully open model flow with training data, checkpoints and recipes for reproducible AI.

Refresh of Grok 4 with stronger reasoning, lower hallucination and faster tool use.

Cost-efficient enterprise model with frontier-class performance for business workloads.

Alibaba's trillion-parameter flagship multilingual reasoning model.

Long-context open agentic model from Moonshot, strong on tool use and coding.

Open bilingual frontier model from Zhipu, competitive on coding and reasoning.

Agentic RL vision-language model for tool-integrated visual reasoning.

AutoQA AI that scores 100% of support conversations across voice and chat.

Speech-to-text, summaries and language operators that analyze every call in real time.

ML-based fraud detection trained on the global Stripe payments network.

AI investing companion delivering market insights to Robinhood Gold customers.

AI suite (DSO.ai, VSO.ai, TSO.ai) optimizing chip design across the EDA flow.

Generative AI for digital chip implementation and verification across the Cadence flow.

Cloud generative-AI app delivering near-instant simulation predictions for engineers.

Tencent's deep-reasoning model, mamba-based and tuned for complex multi-step problems.

Google DeepMind's mathematics model tracked by Epoch, focused on geometry.

In recent months, our focus has been on developing a “good” model while optimizing the developer experience.

We introduce Reka Core, Flash, and Edge, a series of powerful multimodal language models trained from scratch by Reka.

We introduce ChatGLM, an evolving family of large language models that we have been developing over time.

SenseTime's multimodal, language, vision model tracked by Epoch, focused on vision-language generation.

We're contributing Mathstral to the science community to bolster efforts in advanced mathematical problems requiring complex, multi-step logical reasoning.

Modern artificial intelligence (AI) systems are powered by foundation models.

We've developed a new series of AI models designed to spend more time thinking before they respond.

We've developed a new series of AI models designed to spend more time thinking before they respond.

In the past three months since Qwen2’s release, numerous developers have built new models on the Qwen2 language models, providing us with valuable feedback.

In the past three months since Qwen2’s release, numerous developers have built new models on the Qwen2 language models, providing us with valuable feedback.

Qwen2.5 is the latest series of Qwen large language models.

China Telecom's language model tracked by Epoch, focused on language modeling/generation.

NVIDIA's vision, language model tracked by Epoch, focused on language modeling/generation.

NVIDIA's vision, language model tracked by Epoch, focused on language modeling/generation.

NVIDIA's vision, language model tracked by Epoch, focused on language modeling/generation.

Artificial general intelligence start-up Kimi, owned by Chinese AI start-up Moonshot AI, on Saturday launched its first reasoning AI model k0-math.

We've developed a new series of AI models designed to spend more time thinking before they respond.

INTELLECT-MATH is a 7B parameter model optimized for mathematical reasoning.

Language model pretraining with next token prediction has proved effective for scaling compute but is limited to the amount of available training data.

We’re releasing OpenAI o3-mini, the newest, most cost-efficient model in our reasoning series, available in both ChatGPT and the API today.

Tsinghua University's mathematics model tracked by Epoch, focused on mathematical reasoning.

We advance AI capabilities by scaling two complementary paradigms: unsupervised learning and reasoning.

QwQ is the reasoning model of the Qwen series.

As Large Language Models (LLMs) rapidly advance, we introduce Hunyuan-TurboS, a novel large hybrid Transformer-Mamba Mixture of Experts (MoE) model.

We present EXAONE Deep series, which exhibits superior capabilities in various reasoning tasks, including math and coding benchmarks.

In this report, we introduce ERNIE 4.5, a new family of large-scale multimodal models comprising 10 distinct variants.

We've developed a new series of AI models designed to spend more time thinking before they respond.

We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token.

Gemini 2.5 Pro Experimental is our most advanced model for complex tasks.

We’re sharing the first models in the Llama 4 herd, which will enable people to build more personalized multimodal experiences.

Today, we are excited to announce the release of Qwen3, the latest addition to the Qwen family of large language models.

Gemini 2.5 Pro Experimental is our most advanced model for complex tasks.

Claude Sonnet 4 can understand nuanced instructions and context, recognize and correct its own mistakes, and create sophisticated analysis and insights from complex data.

We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1.

Gemini 2.5 Pro Experimental is our most advanced model for complex tasks.

This technical report introduces EXAONE 4.0, which integrates a Non-reasoning mode and a Reasoning mode to achieve both the excellent usability of EXAONE 3.5 and the advanced reasoning abilities of EXAONE Deep.

Today, we are excited to announce the release of Qwen3, the latest addition to the Qwen family of large language models.

Today, we are excited to announce the release of Qwen3, the latest addition to the Qwen family of large language models.

We introduce MindLink, a new family of large language models developed by Kunlun Inc.

To advance Gemini’s capabilities towards solving hard reasoning problems, we developed a novel reasoning approach, called Deep Think, that naturally blends in parallel thinking techniques during response generation.

Today we're releasing Claude Opus 4.1, an upgrade to Claude Opus 4 on agentic tasks, real-world coding, and reasoning.

We introduce LongCat-Flash, a 560-billion-parameter Mixture-of-Experts (MoE) language model designed for both computational efficiency and advanced agentic capabilities.

Large language models (LLMs) have evolved into agentic systems capable of autonomous tool use and multi-step reasoning for complex problem-solving.

Ling-1T is the first flagship non-thinking model in the Ling 2.0 series, featuring 1 trillion total parameters with ≈ 50 billion active parameters per token.

We present Tongyi DeepResearch, an agentic large language model, which is specifically designed for long-horizon, deep information-seeking research tasks.

Today, we are introducing Kimi K2 Thinking, our best open-source thinking model.

Today, we’re excited to launch two powerful new additions to the xAI API: Grok 4.1 Fast, our best tool-calling model with a 2M context window.

DeepSeek's language model tracked by Epoch, focused on mathematical reasoning.

SK Telecom's language model tracked by Epoch, focused on code generation.

K-EXAONE is a large-scale multilingual language model developed by LG AI Research.

Liquid AI's second-generation efficient foundation models built on liquid neural networks for on-device use.

LLM-discovered preference optimization algorithm from Sakana's evolutionary research line.

100M-token context model purpose-built for whole-repository software synthesis.

Code-first foundation model trained with reinforcement learning from code-execution feedback.

Perplexity's in-house search-grounded LLM powering the Perplexity answer engine.

Imbue's research model trained from scratch for robust agentic reasoning and code.

Open-source aligned LLM family known for steerable, uncensored research use.

Alibaba Cloud's closed-source trillion-parameter flagship LLM for coding, reasoning, and enterprise agentic workflows.

Cohere's open-source W4A4-quantized vision-language reasoning model for agentic, multilingual, tool-use enterprise tasks.

Google DeepMind's closed-source natively multimodal reasoning model for fast, high-capability agentic and coding tasks.

NVIDIA's open 14B text-generation LM supporting autoregressive, diffusion-style parallel, and self-speculative decoding.

Sapient's 1B Hierarchical Reasoning Model for compact, structured chain-of-thought text generation.