Comparar modelos
A metric-by-metric comparison using the latest available benchmark, runtime, and pricing values. Tie wins more comparable metrics.
| Metric | Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Kimi K3 |
|---|---|---|
| Artificial Analysis Intelligence Index | 59,9 | 57,1 |
| Artificial Analysis Coding Index | 76,5 | 76,2 |
| LiveBench Mathematics | 96% | 84,4% |
| GPQA | 92,6% | 93,5% |
| Humanity's Last Exam | 53,3% | 44,3% |
| SciCode | 60,2% | 58,7% |
| Output Speed | 70,4 tok/s | 38,4 tok/s |
| Time to First Token | 58,28 s | 5,79 s |
| Blended Price | 20,00 US$ / 1M | 6,00 US$ / 1M |
| Input Price | 10,00 US$ / 1M | 3,00 US$ / 1M |
| Output Price | 50,00 US$ / 1M | 15,00 US$ / 1M |
| Value Index | 3 | 9,52 |
