> Source: LMSYS Chatbot Arena

> Updated: loading...

> arena.lmsys.org

> Click any model row to view stat breakdown

Rank Model Category Score
Loading leaderboard data...

Google

Gemini 2.5 Pro

Flagship model - deep reasoning, multimodal. 1M token context window.

> gemini.google.com

Gemini 2.5 Flash

Fast inference, competitive performance. High-throughput applications.

> ai.google.dev

Anthropic

Claude Opus 4

High-intelligence model for complex reasoning tasks.

> anthropic.com

Claude Sonnet 4

Balanced performance and speed. Coding and analysis specialist.

> anthropic.com

OpenAI

GPT-4o

Multimodal flagship with vision, text, and audio capabilities.

> openai.com

o3

Advanced reasoning model for math, science, and coding.

> openai.com/o3

Meta

Llama 4

Open-weight frontier model family. Research and fine-tuning.

> ai.meta.com

Mistral

Mistral Large

Efficient European AI model. Strong coding and reasoning.

> mistral.ai