APEX-Agents
Long professional tasks for banking, consulting and law agents, graded by experts. Published by Mercor.
| Category | Agents |
|---|---|
| Metric | Accuracy |
| Source | Epoch AI (evaluator) |
| License | CC BY 4.0 |
| Benchmark page | www.mercor.com |
| Status | Scored |
| Weight | 1.0 within Agents |
| Fit | Discrimination 0.93, noise 0.23 |
Results
| # | Model | Provider | Result | Date |
|---|---|---|---|---|
| 1 | Gemini 4 Carbon | 84.2% | 2026-11-11 | |
| 2 | Gemini 4 Argon | 82.4% | 2026-11-11 | |
| 3 | Claude Fable 5.2 | Anthropic | 80.1% | 2026-11-11 |
| 4 | Claude Sonnet 5.5 Max | Anthropic | 75.5% | 2026-11-11 |
| 5 | Claude Opus 5.5 Max | Anthropic | 73.5% | 2026-11-11 |
| 6 | DeepSeek V4.1 Flash 1017 Max | DeepSeek | 69.8% | 2026-11-11 |
| 7 | Claude Fable 5.1 | Anthropic | 68.6% | 2026-11-11 |
| 8 | GPT-6.1 Terra Max | OpenAI | 68.3% | 2026-11-11 |
| 9 | Gemini 3.7 Flash | 67.8% | 2026-11-11 | |
| 10 | Muse Wonder Max | Meta | 66.1% | 2026-11-11 |
| 11 | Claude Opus 5 Max | Anthropic | 65.8% | 2026-11-11 |
| 12 | Kimi K3.1 Pro | Moonshot AI | 65.5% | 2026-11-11 |
| 13 | Grok 4.6 | xAI | 65.3% | 2026-11-11 |
| 14 | GPT-6 Astra | OpenAI | 64.7% | 2026-11-11 |
| 15 | Muse Ember | Meta | 64.3% | 2026-11-11 |
| 16 | Gemini 3.8 Flash | 64.3% | 2026-11-11 | |
| 17 | Claude Fable 5 | Anthropic | 63.6% | 2026-11-11 |
| 18 | Qwen3.8 Max | Alibaba | 63.3% | 2026-11-11 |
| 19 | GPT-6.1 Sol Max | OpenAI | 60.0% | 2026-11-11 |
| 20 | Claude Fable 5.1 High | Anthropic | 59.7% | 2026-11-11 |
| 21 | Muse Spark 1.3 | Meta | 58.6% | 2026-11-11 |
| 22 | GPT-5.6 Terra Max | OpenAI | 58.2% | 2026-11-11 |
| 23 | GLM-5.3 | Zhipu AI | 56.6% | 2026-11-11 |
| 24 | Grok 4.5 | xAI | 56.2% | 2026-11-11 |
| 25 | Claude Haiku 5.5 Max | Anthropic | 56.1% | 2026-11-11 |
| 26 | GPT-5.5 | OpenAI | 55.1% | 2026-11-11 |
| 27 | Grok 4.7 | xAI | 54.6% | 2026-11-11 |
| 28 | Claude Sonnet 5 | Anthropic | 54.5% | 2026-11-11 |
| 29 | MiMo-V2.6-Pro | Xiaomi | 54.4% | 2026-11-11 |
| 30 | GPT-6 Sol Max | OpenAI | 54.3% | 2026-11-11 |
| 31 | GLM-5.3-Flash | Zhipu AI | 52.8% | 2026-11-11 |
| 32 | Claude Opus 5.5 Medium | Anthropic | 52.5% | 2026-11-11 |
| 33 | GPT-5.4 | OpenAI | 52.4% | 2026-11-11 |
| 34 | Mistral Large 4 Preview High | Mistral | 51.6% | 2026-11-11 |
| 35 | GPT-5.6 Sol Pro Max | OpenAI | 51.4% | 2026-11-11 |
| 36 | Claude Opus 4.7 Max | Anthropic | 49.2% | 2026-11-11 |
| 37 | Claude Opus 4.8 Max | Anthropic | 48.9% | 2026-11-11 |
| 38 | Beam 1.0 Max | Reflection | 48.8% | 2026-11-11 |
| 39 | Kimi K3 | Moonshot AI | 47.8% | 2026-11-11 |
| 40 | Muse Spark 1.3 Max | Meta | 47.6% | 2026-11-11 |
| 41 | Qwen3.8 27B | Alibaba | 47.5% | 2026-11-11 |
| 42 | DeepSeek V4 Pro 0813 | DeepSeek | 47.3% | 2026-11-11 |
| 43 | Gemini 3.6 Flash | 46.9% | 2026-11-11 | |
| 44 | Inkling 2 Small | Thinking Machines | 46.8% | 2026-11-11 |
| 45 | Claude Opus 4.6 Max | Anthropic | 46.3% | 2026-11-11 |
| 46 | Muse Spark 1.2 | Meta | 45.4% | 2026-11-11 |
| 47 | GLM-5.2 | Zhipu AI | 45.2% | 2026-11-11 |
| 48 | Claude Sonnet 5.5 Medium | Anthropic | 44.6% | 2026-11-11 |
| 49 | GPT-6 Luna Max | OpenAI | 44.3% | 2026-11-11 |
| 50 | GPT-5.6 Luna | OpenAI | 43.0% | 2026-11-11 |
| 51 | Claude Sonnet 4.6 High | Anthropic | 43.0% | 2026-11-11 |
| 52 | GLM-5.1 | Zhipu AI | 40.9% | 2026-11-11 |
| 53 | DeepSeek V4.1 Flash | DeepSeek | 39.5% | 2026-11-11 |
| 54 | MiniMax-M3 | MiniMax | 37.7% | 2026-11-11 |
| 55 | Kimi K2.7 Code | Moonshot AI | 37.6% | 2026-11-11 |
| 56 | Gemini 3.1 Pro Preview | 35.1% | 2026-11-11 | |
| 57 | Inkling | Thinking Machines | 33.8% | 2026-11-11 |
| 58 | Muse Spark Open | Meta | 33.4% | 2026-11-11 |
| 59 | Muse Spark 1.1 | Meta | 31.8% | 2026-11-11 |
| 60 | Gemini 3.5 Flash Lite | 29.3% | 2026-11-11 | |
| 61 | Gemini 3.5 Flash | 27.5% | 2026-11-11 | |
| 62 | Qwen3.5 397B-A17B | Alibaba | 24.9% | 2026-11-11 |
| 63 | Nemotron 3 Ultra | NVIDIA | 22.7% | 2026-11-11 |
| 64 | DeepSeek V3.2 | DeepSeek | 21.3% | 2026-11-11 |
| 65 | GPT OSS 120B | OpenAI | 4.4% | 2026-11-11 |