AA-Briefcase benchmark
24 AI models ranked on AA-Briefcase. Each row shows the model’s score next to its cheapest API price per 1M tokens where it has one, so you can weigh quality against cost.
Scores undated
Claude Fable 5 (Anthropic) leads at 54%
Models ranked
24
on this benchmark
Top score
54%
on the frozen board
Type
Single eval
benchmark
Frozen
n/a
benchmark scores
Cost vs AA-Briefcase
24 priced modelsEach labelled name links to that model’s page. Curated to the cost-efficiency frontier, top scorers and cheapest so labels stay readable; the full field is in the leaderboard table below (and via “select all”). Reasoning models (extended “thinking” before answering) carry a ring around the dot.
Models ranked on AA-Briefcase
highest score first · 24 models| # | ||||||
|---|---|---|---|---|---|---|
| 1 | Claude Fable 5 |  Anthropic | 54% | $10.00 via Anthropic | $50.00 via Anthropic | |
| 2 | Claude Sonnet 5 | Anthropic | 45% | $2.00 via Anthropic | $10.00 via Anthropic | |
| 3 | Claude Opus 4.8 | Anthropic | 43% | $5.00 via Anthropic | $25.00 via Anthropic | |
| 4 | Claude Opus 4.7 | Anthropic | 40% | $5.00 via Anthropic | $25.00 via Anthropic | |
| 5 | GLM 5.2 |  Zhipu | 38% | $0.05 Routed quote | $1.98 via Scx AI | |
| 6 | GPT-5.5 |  OpenAI | 33% | $5.00 via OpenAI | $30.00 via OpenAI | |
| 7 | MiniMax M3 |  MiniMax | 31% | $0.23 via Wandb | $0.96 via Wandb | |
| 8 | GLM 5.1 | Zhipu | 24% | $1.05 via DeepInfra | $3.50 via DeepInfra | |
| 9 | DeepSeek V4 Pro 0423 |  DeepSeek | 22% | $0.435 via Tencent | $0.87 via Tencent | |
| 10 | MiMo-V2.5-Pro |  Xiaomi | 19% | $0.435 Routed quote | $0.87 Routed quote | |
| 11 | Gemini 3.5 Flash |  Google | 18% | $1.50 via Google | $9.00 via Google | |
| 12 | DeepSeek V4 Flash 0423 | DeepSeek | 16% | $0.09 via DeepInfra | $0.18 via DeepInfra | |
| 13 | Kimi K2.6 |  Moonshot | 16% | $0.65 via Wandb | $3.40 via Novita | |
| 14 | Qwen3.6 27B |  Alibaba | 15% | $0.15 via Libertai | $0.50 via Libertai | |
| 15 | Grok 4.3 |  xAI | 13% | $1.25 via xAI | $2.50 via xAI | |
| 16 | GPT-5.4 Mini | OpenAI | 10% | $0.75 via Databricks | $4.50 via OpenAI | |
| 17 | Qwen3.5 397B A17B | Alibaba | 2% | $0.164 via Aihubmix | $0.986 via Aihubmix | |
| 18 | Mistral Medium 3.5 |  Mistral | 0% | $1.50 via Mistral | $7.50 via Mistral | |
| 19 | gpt-oss-120b | OpenAI | 0% | $0.03 via Wandb | $0.17 via DeepInfra | |
| 20 | gpt-oss-20b | OpenAI | 0% | $0.0145 via Darkbloom | $0.07 via Darkbloom | |
| 21 | Gemini 3.1 Pro Preview | Google | 0% | $2.00 via Google | $12.00 via Google | |
| 22 | Llama 4 Maverick |  Meta | 0% | $0.188 Routed quote | $0.652 Routed quote | |
| 23 | Gemini 3.1 Flash Lite Preview | Google | 0% | $0.25 via Google | $1.50 via Google | |
| 24 | Gemma 4 31B | Google | 0% | $0.14 via Amazon Bedrock | $0.40 via Amazon Bedrock |
Frequently asked questions
What is the AA-Briefcase benchmark?
AA-Briefcase (Artificial Analysis) tests models on realistic, multi-week knowledge-work projects, each with many linked tasks and thousands of fragmented source files such as Slack threads, emails, meeting transcripts and spreadsheets. It combines rubric grading with pairwise quality comparisons to score task success, analytical quality and presentation. The normalised score shown here is demanding: even the strongest models satisfy only a small fraction of tasks in full.
Which AI model scores highest on AA-Briefcase?
Claude Fable 5 (Anthropic) leads with 54%, from $10.00 per 1M input tokens from Anthropic. We cannot tell whether this benchmark is still being published. Treat these scores as undated.
How many models are ranked on AA-Briefcase?
24 models carry an AA-Briefcase score. Scores come from Artificial Analysis; where a model has a listed API price it is drawn from LiteLLM and OpenRouter, and models with no current price are listed without one.
Other benchmarks
compare the same models on a different evalEvery weekday
AI moves fast. Here's your debrief.
News, analysis, tools, and more.
For people who build with AI