gpt-oss-20b is the stronger all-rounder.Nemotron 3 Nano is better at reasoning.
Scores updated · 9 tests both models report · How we compare
Where each one wins
Tests won in each of the three areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
Agents and reasoning rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where gpt-oss-20b pulls ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+27.8points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+18.3points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+14.6points ahead
Where Nemotron 3 Nano pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+5.3points ahead
Every test, side by side
All 9 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Codinggpt-oss-20b
- SWE-bench Multilingualgpt-oss-20b by 27.841.914.1+27.8
- SWE-bench Verifiedgpt-oss-20b by 18.352.434.1+18.3
- SciCodegpt-oss-20b by 8.838.930.1+8.8
- Terminal-Bench 2.1gpt-oss-20b by 5.613.98.3+5.6
Other results3 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- PinchBenchNemotron 3 Nano by 8.957.266.1+8.9
- HLE (text-only, no tools)gpt-oss-20b by 2.913.810.9+2.9
- MMLU-ProNemotron 3 Nano by 2.176.478.5+2.1
Questions people ask
Which is better, gpt-oss-20b or Nemotron 3 Nano?
gpt-oss-20b wins two of the three areas where both have results: coding and agents. Nemotron 3 Nano wins reasoning.
Which is better for coding?
gpt-oss-20b. It wins 4 of the 4 coding tests both models report; Nemotron 3 Nano wins none.
How do you compare the two?
We use the 9 benchmark tests both models have published scores on. The verdict counts the 6 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 3 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.