Measured over the last 30 days
- Models Tracked
- 8
- Avg Tokens / Second
- 19.01
- Avg Time to First Token (ms)
- 6016.25
- Last Updated
- Oct 27, 2026
All dekallm models
8 live · one row per model · fastest first · 1 folded behind their current lane| Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|
| Qwen3 30B A3B Instruct 2507 | 45.70 | 43.80 | 47.70 | 780.00 |
| DeepSeek V4 Pro 0813 | 31.10 | 1.83 | 49.30 | 8920.00 |
| Gemma 4 26B A4B | 29.60 | 20.60 | 44.60 | 790.00 |
| Mistral Nemo | 15.60 | 12.90 | 18.30 | 770.00 |
| Nemotron 3 Super | 9.15 | 9.15 | 9.15 | 6180.00 |
| Qwen3.6 35B A3B | 7.50 | 5.76 | 9.96 | 8710.00 |
| gpt-oss-20b | 6.82 | 1.69 | 18.90 | 13000.00 |
| gpt-oss-120b | 6.57 | 3.78 | 8.82 | 8980.00 |
Frequently Asked Questions
Which dekallm model is fastest?
Based on recent tests, Qwen3 30B A3B Instruct 2507 shows the highest average throughput among tracked dekallm models.
How many recent measurements feed this dashboard?
This provider summary aggregates 984 individual prompts measured across 948 monitoring runs over the past month.