AA-Briefcase Elo
Agents · 2026-06-18
AA-Briefcase is an Artificial Analysis benchmark for long-horizon agentic knowledge-work projects. It evaluates agents on realistic multi-week business workflows with many linked tasks and thousands of input files, combining rubric and pairwise grading into an Elo-style score.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5 | 1470 |
| Sonnet 5 | 1385 |
| Grok 4.5 | 1317 |
| Opus 4.7 | 1280 |
| MiniMax M3 | 1108 |
| GPT-5.5 | 1099 |
| Sonnet 4.6 | 1076 |
| GLM-5.1 | 971 |
| Gemini 3.6 Flash | 962 |
| DeepSeek-V4-Pro | 931 |
| Qwen3.7-Max | 914 |
| MiMo-V2.5-Pro | 879 |
| Nemotron 3 Ultra 550B A55B | 874 |
| Gemini 3.5 Flash | 873 |
| GPT-5.3-Codex | 869 |
| Muse Spark 1.1 | 868 |
| DeepSeek-V4-Flash | 833 |
| Kimi K2.6 | 818 |
| Qwen3.6 27B | 810 |
| Grok 4.3 | 759 |
| GPT-5.4 Mini | 717 |
| Muse Spark | 641 |
| Gemini 3.5 Flash-Lite | 635 |
| Haiku 4.5 | 611 |
| KAT-Coder-Pro V1 | 598 |
| Qwen3.5 397B A17B | 553 |
| Mistral Medium 3.5 | 516 |
| Gemini 3.1 Pro Preview | 457 |
| Gemma 4 31B IT | 373 |
| Command A+ (May 2026, BF16) | 369 |
| North Mini Code | 238 |
| Gemini 3.1 Flash-Lite Preview | 230 |
| Solar Pro 3 | 138 |
| K2 Think V2 | 58.8 |
| gpt-oss-120b | 7.0 |
| gpt-oss-20b | -36.8 |
| Nemotron 3 Super 120B A12B | -95.3 |
| Llama 4 Maverick Instruct | -107 |