Measured over the last 30 days
- Models Tracked
- 11
- Avg Tokens / Second
- 22.47
- Avg Time to First Token (ms)
- 4879.09
- Last Updated
- Oct 17, 2026
All wafer models
11 live · one row per model · fastest first · 6 folded behind their current lane| Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|
| GLM 5.3 | 49.10 | 33.50 | 60.80 | 850.00 |
| DeepSeek V4 Pro 0813 | 44.70 | 1.84 | 74.50 | 1570.00 |
| GLM 5.2 | 39.20 | 9.31 | 67.50 | 1650.00 |
| DeepSeek V4 Flash 0423 | 26.40 | 8.78 | 48.10 | 1310.00 |
| Kimi K3 | 19.80 | 19.80 | 19.80 | 2510.00 |
| Nemotron 3.5 Lightning | 19.10 | 19.10 | 19.10 | 3160.00 |
| DeepSeek V4 Flash 0731 | 18.30 | 2.99 | 55.10 | 5350.00 |
| Switchyard | 11.60 | 11.60 | 11.60 | 1820.00 |
| Qwen3.8 27B | 7.58 | 2.41 | 12.80 | 14100.00 |
| DeepSeek Flash Latest | 5.76 | 5.76 | 5.76 | 9850.00 |
| DeepSeek V4.1 Flash | 5.65 | 5.65 | 5.65 | 11500.00 |
Frequently Asked Questions
Which wafer model is fastest?
Based on recent tests, GLM 5.3 shows the highest average throughput among tracked wafer models.
How many recent measurements feed this dashboard?
This provider summary aggregates 1100 individual prompts measured across 1052 monitoring runs over the past month.