Measured over the last 30 days
- Models Tracked
- 9
- Avg Tokens / Second
- 14.83
- Avg Time to First Token (ms)
- 12227.78
- Last Updated
- Oct 14, 2026
All dekallm models
9 live · one row per model · fastest first · 2 folded behind their current lane| Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|
| Gemma 4 26B A4B | 43.90 | 14.10 | 72.60 | 910.00 |
| Qwen3 30B A3B Instruct 2507 | 43.80 | 43.80 | 43.80 | 810.00 |
| Mistral Nemo | 18.30 | 18.30 | 18.30 | 730.00 |
| Qwen3.6 35B A3B | 8.18 | 5.76 | 9.96 | 8010.00 |
| Qwen3.8 27B | 5.68 | 2.46 | 7.85 | 13300.00 |
| gpt-oss-120b | 5.25 | 4.90 | 5.60 | 9690.00 |
| gpt-oss-20b | 4.27 | 1.69 | 9.44 | 16700.00 |
| Nemotron 3 Super | 2.07 | 2.07 | 2.07 | 29400.00 |
| DeepSeek V4.1 Flash | 2.04 | 2.04 | 2.04 | 30500.00 |
Frequently Asked Questions
Which dekallm model is fastest?
Based on recent tests, Gemma 4 26B A4B shows the highest average throughput among tracked dekallm models.
How many recent measurements feed this dashboard?
This provider summary aggregates 957 individual prompts measured across 925 monitoring runs over the past month.