Muse Spark82.4%
Apr 2026 · Public subsetpublic 500; 100 tool calls
Scale AI ↗Best LLMs for tool use
MCP-AtlasMCP-Atlas evaluates real-world tool use through the Model Context Protocol. It tests tool discovery, choosing correct arguments, recovering from errors, and combining tool outputs to complete multi-step workflows. The stored results below cover the public subset under the April 2026 protocol.
Updated: Data as of:
| Model / creator | Version | Reported score |
|---|---|---|
 Muse SparkMeta · Proprietary | Apr 2026 · Public subset | 82.4% |
 Claude Opus 4.7Anthropic · Proprietary | Apr 2026 · Public subset | 80.8% |
 Gemini 3.1 ProGoogle · Proprietary | Apr 2026 · Public subset | 80.6% |
 GPT-5.5OpenAI · Proprietary | Apr 2026 · Public subset | 79.4% |
| Claude Opus 4.6Anthropic · Proprietary | Apr 2026 · Public subset | 79% |
View all 5 published reports
Claude Opus 4.780.8%
Apr 2026 · Public subsetmax; public 500; 100 tool calls
Scale AI ↗Gemini 3.1 Pro80.6%
Apr 2026 · Public subsethigh; public 500; 100 tool calls
Scale AI ↗GPT-5.579.4%
Apr 2026 · Public subsetxhigh; public 500; 100 tool calls
Scale AI ↗max; public 500; 100 tool calls
Scale AI ↗