# MCP Atlas (original harness) — AI model rankings

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

12 tracked models have a published MCP Atlas (original harness) score. Higher is better. Scores come from published lab reports and benchmark sources; recorded sources appear beside the scores.

## Ranking

| Rank | Model | Developer | Score | Source | Released |
| --- | --- | --- | --- | --- | --- |
| 1 | Gemini 3.1 Pro | Google | 69.2% | Lab | Feb 19 2026 |
| 2 | GPT-5.4 | OpenAI | 67.2% | Lab | Mar 5 2026 |
| 3 | Claude Opus 4.5 | Anthropic | 62.3% | Lab | Nov 24 2025 |
| 4 | Claude Sonnet 4.6 | Anthropic | 61.3% | Lab | Feb 17 2026 |
| 5 | GPT-5.2 | OpenAI | 60.6% | Lab | Dec 11 2025 |
| 6 | Claude Opus 4.6 | Anthropic | 59.5% | Lab | Feb 5 2026 |
| 7 | GPT-5.4 mini | OpenAI | 57.7% | Lab | Mar 17 2026 |
| 8 | Gemini 3.0 Flash | Google | 57.4% | Lab | Dec 17 2025 |
| 9 | GPT-5.4 nano | OpenAI | 56.1% | Lab | Mar 17 2026 |
| 10 | Gemini 3.0 Pro | Google | 54.1% | Lab | Nov 18 2025 |
| 11 | Claude Sonnet 4.5 | Anthropic | 43.8% | Lab | Sep 29 2025 |
| 12 | Claude Opus 4.1 | Anthropic | 40.9% | Lab | Aug 5 2025 |


---

Canonical page: https://aireleasetracker.com/benchmark/mcp-atlas-pre-april-2026
Site index: https://aireleasetracker.com/llms.txt
Source: AI Release Tracker (https://aireleasetracker.com). Most benchmark scores come from lab launch material; gathered results identify the leaderboard that published them.