llm-benchmarks
githubdrose.io

gpt-oss-20b Benchmarks


Measured on groq

5 runs
groq site
Avg Tokens / Second
70.10
Decode Tokens / Second
Avg Time to First Token (ms)
900.00
Runs Analysed
5
Last Updated
Sep 6, 2026, 07:43 AM

Throughput distribution

Throughput over time

gpt-oss-20bgroq310tok/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 · models with at least 4 samples in the window first, fastest of those at the top · shared vertical scale, 0 to 391 tok/s (99th percentile) · dashed rule is the model's own mean

Every lane measuring this model

1 measured · live first · a lane is a provider, a catalogue id and a transport
ProviderLaneAvg Toks/SecMinMaxAvg TTF (ms)n
groqvia OpenRouter70.1051.40110.00900.005

Frequently Asked Questions

How fast is gpt-oss-20b?
The latest rolling average throughput is 70.10 tokens per second with an average time to first token of 900.00 ms across 5 recent runs.
How often are these benchmarks updated?
Benchmarks refresh automatically whenever the monitoring cron runs. The most recent run completed on Sep 6, 2026, 07:43 AM.

Related models