Benchmarks · agentic
Mercor APEX
Score: best pass@1 (loop_truncated_tools_agent). Collected by Jake Boggs' benchmark dashboard.
61 models · 61 results · data as of 2026-11-06 · mercor.com ↗
Leaderboard
Gemini 4 Carbon max
All results
61 rows| # | Model | Effort | Mercor APEX |
|---|---|---|---|
| 1 | Gemini 4 Carbon | max | 84.2% |
| 2 | Gemini 4 Argon | max | 82.4% |
| 3 | Sonnet 5.5 | max | 75.5% |
| 4 | Opus 5.5 | max | 73.5% |
| 5 | DeepSeek V4.1 Flash 1017 | max | 69.8% |
| 6 | Fable 5.1 | max | 68.6% |
| 7 | GPT-6.1 Terra | max | 68.3% |
| 8 | Gemini 3.7 Flash | max | 67.8% |
| 9 | Muse Wonder (Max) | max | 66.1% |
| 10 | Opus 5 | max | 65.8% |
| 11 | Kimi K3.1 | max | 65.5% |
| 12 | Grok 4.6 | max | 65.3% |
| 13 | GPT-6 Astra | max | 64.7% |
| 14 | Gemini 3.8 Flash | max | 64.3% |
| 15 | Fable 5 | max | 63.6% |
| 16 | Qwen 3.8 Max | max | 63.3% |
| 17 | GPT-6.1 Sol | max | 60.0% |
| 18 | Muse Spark 1.3 | max | 58.6% |
| 19 | GPT-5.6 Terra | max | 58.2% |
| 20 | MiMo V2.6 Flash | high | 57.4% |
| 21 | GLM 5.3 | max | 56.6% |
| 22 | Grok 4.5 | max | 56.2% |
| 23 | Haiku 5.5 | max | 56.1% |
| 24 | DeepSeek V4 Flash | max | 55.3% |
| 25 | GPT-5.5 | xhigh | 55.1% |
| 26 | Grok 4.7 | max | 54.6% |
| 27 | Sonnet 5 | max | 54.5% |
| 28 | MiMo V2.6 Pro | max | 54.4% |
| 29 | GPT-6 Sol | max | 54.3% |
| 30 | MiniMax M3.1 | max | 53.9% |