- Avg Tokens / Second
- 21.15
- Avg Time to First Token (ms)
- 835.00
- Runs Analysed
- 67
- Last Updated
- Sep 2, 2026, 04:07 AM
Throughput distribution
Throughput over time
Generated throughput, including reasoning tokens — so a thinking model's line does not collapse; the table ranks on visible tokens · one cell per model, showing its best-covered provider · shared vertical scale, 0 to 25 tok/s (99th percentile) · dashed rule is the model's own mean · 12 further providers not drawn · 1 samples above the ceiling drawn at it
Every provider serving this model
2 measured| Provider | Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|---|
| deepinfra | Qwen3 235B A22B Instruct 2507 | 21.40 | 15.50 | 25.60 | 720.00 |
| deepinfra | Qwen3 235B A22B Instruct 2507 | 20.90 | 6.33 | 34.50 | 950.00 |
Frequently Asked Questions
How fast is Qwen3 235B A22B Instruct 2507?
The latest rolling average throughput is 21.15 tokens per second with an average time to first token of 835.00 ms across 67 recent runs.
How often are these benchmarks updated?
Benchmarks refresh automatically whenever the monitoring cron runs. The most recent run completed on Sep 2, 2026, 04:07 AM.