LLM Model Directory

Documentation links and comparison for 10 providers and 20 latest models. Quick access to docs, API references, and model specs.

OpenAIProprietary

Creators of ChatGPT and the GPT series. Leading commercial AI lab.

GPT-4.1

Best for codingInstruction followingLong context

1M tokensApr 2025

GPT-4o

MultimodalFastAudio/Vision

128K tokensMay 2024

Reasoning modelChain-of-thoughtSTEM & code

200K tokensApr 2025

o4-mini

Fast reasoningCost-efficientTool use

200K tokensApr 2025

AnthropicProprietary

Makers of Claude. Focus on AI safety and helpfulness.

Claude Opus 4.6

Most capableDeep reasoningComplex tasks

1M tokensMar 2025

Claude Sonnet 4.6

Fast & capableGreat for codingBalanced

200K tokensMar 2025

Claude Haiku 4.5

FastestCost-effectiveLightweight tasks

200K tokensOct 2025

Google DeepMindProprietary

Gemini model family powering Google AI products.

Gemini 2.5 Pro

Thinking modelTop benchmarksMultimodal

1M tokensMar 2025

Gemini 2.5 Flash

FastCost-efficientThinking toggle

1M tokensMar 2025

Meta (Llama)Open Source

Open-weight LLMs from Meta AI. Free for research and commercial use.

Llama 4 Maverick

400B MoE17 active expertsMultimodal

1M tokensApr 2025

Llama 4 Scout

109B MoE16 expertsLongest context

10M tokensApr 2025

Mistral AIOpen Source

European AI lab building efficient, open-weight models.

Mistral Large

Flagship modelMultilingualFunction calling

128K tokensMar 2025

Codestral

Code-specialized80+ languagesFIM support

256K tokensJan 2025

DeepSeekOpen Source

Chinese AI lab known for efficient, high-performing open-source models.

DeepSeek V3

685B MoETop codingOpen weights

128K tokensMar 2025

DeepSeek R1

Reasoning modelChain-of-thoughtMath & logic

128K tokensJan 2025

xAI (Grok)Proprietary

Elon Musk's AI company. Grok models with real-time knowledge.

Grok 3

Real-time infoReasoningMultimodal

128K tokensFeb 2025

Alibaba (Qwen)Open Source

Open-source model family from Alibaba Cloud. Strong multilingual support.

Qwen 2.5

72B paramsMultilingualCode & math

128K tokensJan 2025

QwQ 32B

Reasoning model32B paramsOpen weights

128K tokensMar 2025

Microsoft (Phi)Open Source

Small language models that punch above their weight.

Phi-4

14B paramsReasoningSmall but capable

16K tokensDec 2024

CohereProprietary

Enterprise-focused AI with strong RAG and embedding capabilities.

Command R+

RAG-optimizedEnterpriseTool use

128K tokensMar 2025