Which AI model scores highest on MCP Atlas?

Gemini 3.1 Pro by Google holds the best MCP Atlas (original harness) result among tracked models, at 69.2% (released Feb 19 2026). Higher scores are better on this benchmark.

What are the top 5 models on MCP Atlas?

1. Gemini 3.1 Pro (Google) — 69.2%; 2. GPT-5.4 (OpenAI) — 67.2%; 3. Claude Opus 4.5 (Anthropic) — 62.3%; 4. Claude Sonnet 4.6 (Anthropic) — 61.3%; 5. GPT-5.2 (OpenAI) — 60.6%.

How many models have a published MCP Atlas score?

12 tracked models have a published MCP Atlas (original harness) score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.

Multi-step tool use

MCP Atlasoriginal harness

pass rate under the original tool-use harness, before Scale changed the judge and retry handling in April 2026

MCP Atlas original harness

Rankings

Higher is better

MCP Atlas — frequently asked questions

: pass rate under the original tool-use harness, before Scale changed the judge and retry handling in April 2026
: Gemini 3.1 Pro by Google holds the best MCP Atlas (original harness) result among tracked models, at 69.2% (released Feb 19 2026). Higher scores are better on this benchmark.
: 1. Gemini 3.1 Pro (Google) — 69.2%; 2. GPT-5.4 (OpenAI) — 67.2%; 3. Claude Opus 4.5 (Anthropic) — 62.3%; 4. Claude Sonnet 4.6 (Anthropic) — 61.3%; 5. GPT-5.2 (OpenAI) — 60.6%.
: 12 tracked models have a published MCP Atlas (original harness) score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.