PUBLIC BENCHMARKS. A CLEARER PICTURE.
Compare AI models
across real-world tasks.
Published results, transparent sources.
A clearer view of what different models can do.
GPQA Diamond · reported accuracy
Ranking evidence
| # | USD / 1M tokens | Modalities | Compare | |||||
|---|---|---|---|---|---|---|---|---|
| 1 | Qwen | Input$0.5Output$3.6 | InputTextImageVideoOutputText | 262,144tokens | ||||
| 2 | moonshotai | Input$0.5Output$2.5 | InputTextImageOutputText | 262,144tokens | ||||
| 3 | Qwen | Input$0.26Output$2.08 | InputTextImageVideoOutputText | 262,144tokens | ||||
| 4 | zai-org | — | Input$1Output$3.2 | InputTextOutputText | 204,800tokens | |||
| 5 | zai-org | Input$0.6Output$2.2 | InputTextOutputText | 204,800tokens | ||||
| 6 | Qwen | Input$0.26Output$2.6 | InputTextImageVideoOutputText | 262,144tokens | ||||
| 7 | MiniMaxAI | — | Input$0.27Output$0.95 | InputTextOutputText | 204,800tokens | |||
| 8 | moonshotai | Input$0.6Output$2.5 | InputTextOutputText | 262,144tokens | ||||
| 9 | Qwen | Input$0.15Output$1 | InputTextImageVideoOutputText | 262,144tokens | ||||
| 10 | Nanbeige | — | —Not documented | InputTextOutputNot documented | —Not documented |
Scores share a published source cohort. Evaluation settings are incomplete upstream; interpret differences cautiously. Missing data is shown as a dash.
Evidence at a glance
42 use-case ranking entries added: searchArena · official dataset
165 use-case ranking entries added: visionArena · official dataset
140 use-case ranking entries added: codingArena · official dataset
48 use-case ranking entries added: videoArena · official dataset