Performance Benchmarks for Kimi K2 Preview (0905)
Analyze how Kimi K2 Preview (0905) stacks up against other frontier models in terms of cost-to-performance and raw intelligence (LMArena Elo).
2026 Performance Summary
- 🚀 Intelligence (Elo): 1418
- 💰 API Cost (Input): $6e-7 / 1M tokens
- ⚡ API Cost (Output): $0.0000025 / 1M tokens
- ⏱️ Real-world Throughput: 35 tokens/s
Analysis: While Kimi K2 Preview (0905) is a capable model, it is currently outperformed in efficiency by Claude Fable 5.2 (xHigh). Claude Fable 5.2 (xHigh) sits higher on the Pareto line, offering approximately 408 points more intelligence (Elo) for a comparable or lower API cost.
Highly Recommended Alternatives
*These models represent the Pareto Frontier (optimal cost-to-performance).*