Compare
Compare by Provider
Two to four providers, ranked side by side. Every model each one publishes lands on one capability table and one quality-versus-cost plot, and a figure no source published stays explicitly unavailable.
Provider set
Check two to 4 providers below, then press Compare — the URL and the table below only update once you do.
Compare by Provider
32 models across 2 providers, on the same seven capability axes the leaderboard ranks. Choose a column header to sort; expand a row for its benchmark tasks.
- Models
- 7
- Scored
- 7
- Best overall
- 74.43Kimi K3
- Blended $ / 1M
- $1.08 – $6.00
- Models
- 25
- Scored
- 25
- Best overall
- 73.17Qwen3.8 Max
- Blended $ / 1M
- $0.055 – $1.56
Evidence key: ObservedUnavailable
| Compare | Details | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 74.43 | 91 | 71.7 | 84.5 | 78.5 | 89.4 | Unavailable | Unavailable | ||
| 2 | 73.17 | 83.6 | 68 | 69 | 86.6 | 87.4 | 90.5 | Unavailable | ||
| 3 | 66.87 | 63.8 | 69.5 | 71 | 75.1 | Unavailable | 89.2 | 81.9 | ||
| 4 | 65.6 | 71 | 27.6 | 65.6 | 75.5 | Unavailable | 75.1 | Unavailable | ||
| 5 | 65.5 | 60.8 | 58.2 | 57.6 | Unavailable | 63.9 | Unavailable | 71 | ||
| 6 | 64.29 | 82.4 | 55.6 | 47.6 | 77.4 | 79.8 | 83.2 | Unavailable | ||
| 7 | 63.94 | 56 | 43.4 | 69.9 | Unavailable | Unavailable | Unavailable | 41.3 | ||
| 8 | 61.6 | 58.1 | 57.9 | 60.1 | 74 | 72.4 | 89.2 | 78.2 | ||
| 9 | 60.36 | 52.5 | 52.6 | 58.5 | 60.1 | 66.2 | 82.5 | 62 | ||
| 10 | 57.54 | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | ||
| 11 | 57.39 | 28 | 58.9 | 81.4 | Unavailable | 63.3 | Unavailable | Unavailable | ||
| 12 | 56.98 | 41 | 54.9 | 53.3 | 75.8 | 83.1 | 87.2 | Unavailable | ||
| 13 | 56.19 | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | 28.4 | ||
| 14 | 56.08 | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | 30 | ||
| 15 | 55.14 | 15.5 | 57 | 71.7 | 51.6 | 70.6 | 91.5 | Unavailable | ||
| 16 | 55.09 | 30.9 | 34.8 | 54.5 | 59.8 | 62.9 | Unavailable | 73.5 | ||
| 17 | 54.11 | Unavailable | Unavailable | 72.1 | Unavailable | Unavailable | Unavailable | Unavailable | ||
| 18 | 54.02 | 28 | 42.8 | 56 | 54.1 | 65.8 | 84.4 | 62.1 | ||
| 19 | 53.47 | 13.7 | 47.6 | 70.5 | 40.2 | 67.2 | 87.4 | Unavailable | ||
| 20 | 50.67 | 24.9 | 51.2 | 73.3 | 48.7 | 57 | 91.6 | Unavailable | ||
| 21 | 50.33 | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | 39.3 | ||
| 22 | 49.46 | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | ||
| 23 | 48.09 | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | ||
| 24 | 47.73 | 47.2 | 41.4 | 46.9 | 74.1 | 51.9 | 81 | 72.1 | ||
| 25 | 43.85 | 36 | 25.6 | 43.3 | 70.3 | 50 | 76.8 | 70.7 | ||
| 26 | 43.1 | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | ||
| 27 | 40.75 | Unavailable | Unavailable | 49.7 | 55.4 | Unavailable | 50.3 | Unavailable | ||
| 28 | 38.64 | Unavailable | Unavailable | 39.2 | 21 | 41.9 | 33.5 | Unavailable | ||
| 29 | 38.27 | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | ||
| 30 | 37.52 | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | ||
| 31 | 33.3 | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | Unavailable | ||
| 32 | 23.12 | Unavailable | Unavailable | 52.9 | 57.4 | Unavailable | 47 | 39 |
12 of these rows are marked self-hosted by the source: open-weight models with no first-party hosted token rate. Their per-token price is unavailable rather than zero.
Quality versus cost
Each point is one of these providers' models with both an overall score and complete token prices. Cost is logarithmic by default; untick for a linear axis.
Lowest blended cost
Up to twelve of these providers' models, ranked on the 75% input / 25% output blend.
12 of these rows are marked self-hosted by the source: open-weight models with no first-party hosted token rate. Their per-token price is unavailable rather than zero.
12 self-hosted models are left out: the source publishes no hosted token rate for them, so they have no comparable per-token cost — not a cost of zero.
0 of 4 selected