Reported by others
Grading short stories (Lech Mazur Writing)
The model writes a short story that must use every required element, such as a character, an object and a setting, and several grader models score it.
How to read this table
- Higher rows are rated higher. A range like “1–2” means the gap is within the margin of error.
- The score is the value published by the task's authors and other organizations.
- Conditions can differ between models, so small gaps do not mean much.
- Prices are a guide per million tokens, reading then writing.
- Length is how much text the model can read at once (tokens).
| Rank | Model | Developer | Score | Input $ /1M | Output $ /1M | Context | Supports | Weights | Released | Price and specs |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | GPT-5 | OpenAI | Rating 86.0 | $1.25 | $10 | 400K | OpenAI Released 2025-08 Price $1.25 / $10 Length 400K | |||
| 2 | Kimi K2 (instruct) | Moonshot | Rating 85.6 | — | — | 131.1K | Moonshot Released 2025-07 Price — Length 131.1K | |||
| 3 | Claude Opus 4.1 | Anthropic | Rating 84.7 | — | — | 200K | Anthropic Released 2025-08 Price — Length 200K | |||
| 4 | o3-pro | OpenAI | Rating 84.4 | — | — | 200K | OpenAI Released 2025-06 Price — Length 200K | |||
| 5 | o3 | OpenAI | Rating 83.9 | $2 | $8 | 200K | OpenAI Released 2025-04 Price $2 / $8 Length 200K | |||
| 6 | Gemini 2.5 Pro | Rating 83.8 | $1.25 | $10 | 1M | Google Released 2025-06 Price $1.25 / $10 Length 1M | ||||
| 7 | Claude Opus 4 | Anthropic | Rating 83.6 | — | — | 200K | Anthropic Released 2025-05 Price — Length 200K | |||
| 8 | GPT-5 mini | OpenAI | Rating 83.1 | $0.25 | $2 | 400K | OpenAI Released 2025-08 Price $0.25 / $2 Length 400K | |||
| 9–10 | DeepSeek-R1 | DeepSeek | Rating 83.0 | — | — | 128K | DeepSeek Released 2025-01 Price — Length 128K | |||
| 9–10 | Qwen3 235B A22B | Alibaba | Rating 83.0 | $0.7 | $2.8 | 128K | Alibaba Released 2025-04 Price $0.7 / $2.8 Length 128K | |||
| 11 | Qwen3 235B A22B (thinking) | Alibaba | Rating 82.4 | — | — | 262.1K | Alibaba Released 2025-07 Price — Length 262.1K | |||
| 12 | DeepSeek-R1 (2025-05) | DeepSeek | Rating 81.9 | — | — | 131.1K | DeepSeek Released 2025-05 Price — Length 131.1K | |||
| 13 | GPT-4o (2024-11) | OpenAI | Rating 81.8 | $2.5 | $10 | 128K | OpenAI Released 2024-11 Price $2.5 / $10 Length 128K | |||
| 14 | Claude Sonnet 4 | Anthropic | Rating 81.4 | — | — | 200K | Anthropic Released 2025-05 Price — Length 200K | |||
| 15 | Claude 3.7 Sonnet | Anthropic | Rating 81.1 | — | — | 200K | Anthropic Released 2025-02 Price — Length 200K | |||
| 16 | Gemini 2.5 Pro (preview, 2025-05) | Rating 80.9 | — | — | 1.05M | Google Released 2025-05 Price — Length 1.05M | ||||
| 17 | Gemini 2.5 Pro (exp) | Rating 80.5 | — | — | 1.05M | Google Released 2025-03 Price — Length 1.05M | ||||
| 18 | Claude 3.5 Sonnet (2024-10) | Anthropic | Rating 80.3 | — | — | 200K | Anthropic Released 2024-10 Price — Length 200K | |||
| 19 | QwQ 32B | Alibaba | Rating 80.2 | — | — | 32.8K | Alibaba Released 2025-03 Price — Length 32.8K | |||
| 20 | Gemma 3 27B | Rating 79.9 | — | — | 128K | Google Released 2025-03 Price — Length 128K | ||||
| 21 | Mistral Medium 3 | Mistral | Rating 77.3 | $0.4 | $2 | 128K | Mistral Released 2025-05 Price $0.4 / $2 Length 128K | |||
| 22 | GPT-OSS 120B | OpenAI | Rating 77.1 | — | — | 128K | OpenAI Released 2025-08 Price — Length 128K | |||
| 23 | DeepSeek-V3 (2025-03) | DeepSeek | Rating 77.0 | — | — | 163.8K | DeepSeek Released 2025-03 Price — Length 163.8K | |||
| 24 | Grok 4 | xAI | Rating 76.9 | — | — | 256K | xAI Released 2025-07 Price — Length 256K | |||
| 25 | Gemini 2.5 Flash (preview, 2025-04) | Rating 76.5 | — | — | 1.05M | Google Released 2025-04 Price — Length 1.05M | ||||
| 26 | Grok 3 (beta) | xAI | Rating 76.4 | — | — | — | xAI Released 2025-04 Price — Length — | |||
| 27 | GPT-4.5 (preview) | OpenAI | Rating 75.6 | — | — | — | OpenAI Released 2025-02 Price — Length — | |||
| 28 | Qwen3 30B A3B | Alibaba | Rating 75.3 | — | — | 128K | Alibaba Released 2025-04 Price — Length 128K | |||
| 29 | o4-mini | OpenAI | Rating 75.0 | $1.1 | $4.4 | 200K | OpenAI Released 2025-04 Price $1.1 / $4.4 Length 200K | |||
| 30 | Gemini 2.0 Flash (thinking, exp) | Rating 73.8 | — | — | — | Google Released 2024-12 Price — Length — | ||||
| 31–32 | Claude 3.5 Haiku | Anthropic | Rating 73.5 | — | — | 200K | Anthropic Released 2024-10 Price — Length 200K | |||
| 31–32 | Grok 3 Mini (beta) | xAI | Rating 73.5 | — | — | — | xAI Released 2025-04 Price — Length — | |||
| 33 | GLM-4.5 | Z.ai | Rating 73.4 | $0.6 | $2.2 | 131.1K | Z.ai Released 2025-08 Price $0.6 / $2.2 Length 131.1K | |||
| 34 | Qwen2.5 Max | Alibaba | Rating 72.9 | — | — | — | Alibaba Released 2025-01 Price — Length — | |||
| 35 | o1 | OpenAI | Rating 70.2 | $15 | $60 | 128K | OpenAI Released 2024-12 Price $15 / $60 Length 128K | |||
| 36 | Mistral Large 2 | Mistral | Rating 69.0 | — | — | 131.1K | Mistral Released 2024-07 Price — Length 131.1K | |||
| 37 | GPT-4o mini | OpenAI | Rating 67.2 | — | — | 128K | OpenAI Released 2024-07 Price — Length 128K | |||
| 38 | o1-mini | OpenAI | Rating 64.9 | — | — | 128K | OpenAI Released 2024-09 Price — Length 128K | |||
| 39 | Grok 2 (2024-12) | xAI | Rating 63.6 | — | — | — | xAI Released 2024-12 Price — Length — | |||
| 40 | Phi-4 | Microsoft | Rating 62.6 | — | — | 128K | Microsoft Released 2024-12 Price — Length 128K | |||
| 41 | Llama 4 Maverick | Meta | Rating 62.0 | — | — | 524.3K | Meta Released 2025-04 Price — Length 524.3K | |||
| 42 | o3-mini | OpenAI | Rating 61.7 | $1.1 | $4.4 | 200K | OpenAI Released 2025-01 Price $1.1 / $4.4 Length 200K | |||
| 43 | Amazon Nova Pro 1.0 | Amazon | Rating 60.5 | — | — | 300K | Amazon Released 2024-12 Price — Length 300K |
Best score over time
2024-07-18Release date2025-08-07
How to read this page
- The value is the 10-point average rescaled to 100. Figures published by the organizations that made the tasks, among others, collected here.
- The average grade rescaled to 100, and most models score 60 to 90. No error margins are published, so differences of 1 to 2 points are not clear differences.
- In the Supports column the icons are, from left: can use tools / can read images / can answer in a fixed format (JSON); faded means not supported or unknown.
- Open-weights models have their contents (the weights) published, so you can also run them on your own computer or server. Models without the open-weights mark are normally used through the developer's own services (apps or API); for some models the weights status is unknown.
- Names such as ChatGPT next to a model are the consumer services its developer offers. Check on the service itself whether that model is available there.
Data: Arena, Epoch AI, Swallow LLM team (CC BY 4.0), Models.dev (MIT), updated 2026-10-24, data sources