Overall Preference vs Speed
Top-left quadrant shows models with fast generation and high ratings
Overall Preference vs Price
Top-left quadrant shows models with low cost and high ratings
Top-left quadrant shows models with fast generation and high ratings
Top-left quadrant shows models with low cost and high ratings
Top-left quadrant shows models with fast generation and high ratings
Top-left quadrant shows models with low cost and high ratings
One ranking across the HTML and agentic boards.
Claude Opus 5.5
(high)
GPT-6.1 Sol
(max)
GPT-6 Astra
(max)
Claude Fable 5.1
Gemini 4 Argon
Kimi K3.1 Pro
Muse Spark 1.3 Max
Claude Sonnet 5.5
Claude Fable 5
Claude Opus 5
MiMo-V2.6-Pro
Muse Spark 1.3
(xhigh)
Mistral Medium 3.5
Evaluates model ability across a composite of agentic tasks: Frontend Web App, Full Stack, Agentic Game Dev, Godot, React Native, Android, Slides, and Spreadsheet. Each output is built by a coding agent across multiple files and tool calls. Results aggregate real-world user preferences across these categories to provide an overall view of agentic performance.
Top-left quadrant shows models with fast generation and high ratings
Top-left quadrant shows models with low cost and high ratings
Elo Rating: A skill-based rating system starting at 1200. Ratings are calculated using the standard Bradley-Terry model. See more in our Methodology page.Margin of Error: Shows ±standard error from the Bradley-Terry model for Elo rating confidence intervals.