Measured over the last 30 days
- Models Tracked
- 14
- Avg Tokens / Second
- 12.40
- Avg Time to First Token (ms)
- 14424.29
- Last Updated
- Oct 17, 2026
All phala models
14 live · one row per model · fastest first| Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|
| Qwen2.5 7B Instruct | 62.30 | 62.30 | 62.30 | 660.00 |
| GLM 5.2 | 21.10 | 21.10 | 21.10 | 2020.00 |
| gpt-oss-120b | 19.40 | 4.26 | 29.50 | 6010.00 |
| DeepSeek V3.2 | 12.80 | 5.49 | 16.80 | 2630.00 |
| DeepSeek V4 Flash 0731 | 12.10 | 5.22 | 18.90 | 7240.00 |
| DeepSeek V4.1 Flash | 8.68 | 8.68 | 8.68 | 6800.00 |
| Muse Glimmer 30B | 8.61 | 8.61 | 8.61 | 6510.00 |
| Nemotron 3.5 Lightning | 8.55 | 8.55 | 8.55 | 6870.00 |
| Kimi K2.7 Code | 5.49 | 5.49 | 5.49 | 8000.00 |
| Qwen3.6 35B A3B | 5.28 | 0.93 | 10.50 | 18300.00 |
| Hy3 | 2.78 | 1.86 | 3.55 | 24300.00 |
| Kimi K2.6 | 2.66 | 1.53 | 3.54 | 25500.00 |
| Qwen3.5 397B A17B | 2.51 | 1.55 | 3.34 | 27400.00 |
| GLM 5.1 | 1.33 | 0.61 | 2.70 | 59700.00 |
Frequently Asked Questions
Which phala model is fastest?
Based on recent tests, Qwen2.5 7B Instruct shows the highest average throughput among tracked phala models.
How many recent measurements feed this dashboard?
This provider summary aggregates 2710 individual prompts measured across 2402 monitoring runs over the past month.