llm-benchmarks
githubdrose.io

Qwen2.5 VL 72B Instruct Benchmarks


Measured on parasail

43 runs
Avg Tokens / Second
29.95
Avg Time to First Token (ms)
960.00
Runs Analysed
43
Last Updated
Sep 1, 2026, 04:07 AM

Throughput distribution

Throughput over time

Qwen2.5 VL 72B Instructnebius1tok/s

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 31 tok/s (99th percentile) · dashed rule is the model's own mean · 3 further providers not drawn · 1 samples above the ceiling drawn at it

Every provider serving this model

2 measured
ProviderModelAvg Toks/SecMinMaxAvg TTF (ms)
parasailQwen2.5 VL 72B Instruct29.4015.8039.90630.00
parasailQwen2.5 VL 72B Instruct30.509.3040.701290.00

Frequently Asked Questions

How fast is Qwen2.5 VL 72B Instruct?
The latest rolling average throughput is 29.95 tokens per second with an average time to first token of 960.00 ms across 43 recent runs.
How often are these benchmarks updated?
Benchmarks refresh automatically whenever the monitoring cron runs. The most recent run completed on Sep 1, 2026, 04:07 AM.

Related models