MultiChallenge
Chat & Writing · 2025-01-29
Score on realistic multi-turn conversations across retention, memory, editing, and self-coherence.
Top models (higher is better)
| Model | Score |
|---|---|
| Muse Spark | 75.5 |
| Muse Spark 1.1 | 75.3 |
| Gemini 3.1 Pro Preview | 71.4 |
| GPT-5.4 Pro | 69.2 |
| Gemini 3 Pro Preview | 65.7 |
| GPT-5.1 | 63.4 |
| GPT-5 | 63.2 |
| o3-pro | 62.4 |
| Kimi K2.5 | 61.4 |
| Gemini 3.1 Flash-Lite Preview | 60.6 |
| GPT-5 Mini | 59.0 |
| Opus 4.5 | 59.0 |
| Opus 4 | 58.6 |
| Opus 4.1 | 57.2 |
| Sonnet 4 | 57.1 |
| o3 | 56.6 |
| Opus 4.6 | 56.0 |
| Kimi K2 Thinking | 55.4 |
| Sonnet 4.5 | 55.3 |
| Gemini 2.5 Pro | 53.6 |
| Claude 3.7 Sonnet | 51.6 |
| GPT-5.1 Instant | 51.2 |
| Haiku 4.5 | 50.5 |
| DeepSeek-V3.1 | 46.1 |
| gpt-oss-120b | 45.3 |
| O4 Mini | 44.9 |
| Qwen3-235B-A22B | 41.2 |
| GPT-4.1 | 39.4 |
| Gemini 2.0 Flash 001 | 36.4 |