Kimi K2 Turbo (Thinking): benchmark results
Kimi K2 Turbo evaluated with thinking enabled. Provider: Moonshot. Released 2025-11-06. Access: Open.
Unified ELO 1631 ± 1, rank #314 of 1761 rated models, from 26 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Chatbot Arena (Text - Math) | 1437 | Arena Score | 78.4 |
| Chatbot Arena (Text - Coding) | 1486 | Arena Score | 78 |
| Chatbot Arena (Text - English) | 1443 | Arena Score | 76.4 |
| Chatbot Arena (Text - Chinese) | 1474 | Arena Score | 76.1 |
| Chatbot Arena (Text - Hard Prompts) | 1453 | Arena Score | 76.1 |
| Chatbot Arena (Text - Instruction Following) | 1418 | Arena Score | 75.2 |
| Chatbot Arena (Text) | 1430 | Elo | 75.2 |
| Chatbot Arena (Text - Multi-Turn) | 1433 | Arena Score | 74.9 |
| Opper TaskBench | 90.8 | Avg Task Score (%) | 72.9 |
| Chatbot Arena (Text - Creative Writing) | 1392 | Arena Score | 72.4 |
| Chatbot Arena (Text - Longer Query) | 1433 | Arena Score | 71.2 |
| Chatbot Arena (Text - Russian) | 1412 | Arena Score | 69.7 |
Interactive version: theaggregate.ai/model?slug=kimi-k2-turbo-thinking · How It Works · Data refreshed daily, snapshot 2026-09-05.