Model comparison
Amazon Nova Pro vs gpt-oss-20b
gpt-oss-20b is the stronger model overall, scoring 32.5 to 31.0 on the Noometry Index.
Last verified . 22 shared benchmarks.
Summary
- They share 22 benchmarks with published results for both. Amazon Nova Pro scores higher in 5 categories and gpt-oss-20b in 4 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where gpt-oss-20b leads 39.4 to 28.5.
- The biggest single-benchmark swing is Omni-MATH: 24.2% for Amazon Nova Pro and 56.5% for gpt-oss-20b.
- gpt-oss-20b is cheaper at $0.018 / $0.09 per million input/output tokens, against $0.80 / $3.20 for Amazon Nova Pro.
- Amazon Nova Pro accepts more context: 300K tokens versus 131K.
- gpt-oss-20b has downloadable open weights; the other is API-only.
Side by side
| Amazon Nova Pro | gpt-oss-20b | |
|---|---|---|
| Provider | Amazon | OpenAI |
| Noometry Index | 31.0 | 32.5 |
| Released | 2024-12-03 | 2025-08-05 |
| Weights | Proprietary | Open |
| Context window | 300K | 131K |
| Max output | 10K | 16K |
| Input $ / M tokens | $0.80 | $0.018 |
| Output $ / M tokens | $3.20 | $0.09 |
| Results tracked | 38 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding gpt-oss-20b leads
Amazon Nova Pro: 35.1 (#229), gpt-oss-20b: 37.6 (#192)
| Benchmark | Amazon Nova Pro | gpt-oss-20b |
|---|---|---|
| LMArena Coding | 1270 | 1306 |
| SciCode | — | 34.4% |
| WeirdML | — | 40.9% |
| LiveBench Coding | 38.1% | — |
| ALE-Bench | — | 566.05 |
Agentic & Tool Use Amazon Nova Pro leads
Amazon Nova Pro: 16.7 (#147), gpt-oss-20b: 9.3 (#154)
| Benchmark | Amazon Nova Pro | gpt-oss-20b |
|---|---|---|
| Terminal-Bench | — | 3.4% |
| Berkeley Function Calling Leaderboard | 25% | — |
| TheAgentCompany | 1.7% | — |
Reasoning Too close to call
Amazon Nova Pro: 20.0 (#243), gpt-oss-20b: 19.3 (#261)
| Benchmark | Amazon Nova Pro | gpt-oss-20b |
|---|---|---|
| LMArena Hard Prompts | 1246 | 1274 |
| Epoch Capabilities Index | 123.8 | 137.82 |
| Kagi LLM Benchmark | — | 53.2% |
| CritPt | — | 1.4% |
| Chess Puzzles | — | 4% |
| LiveBench Reasoning | 32.6% | — |
| DTBench | — | 68% |
| LiveBench Data Analysis | 48.3% | — |
| LMCA | — | 14.5% |
| LiveBench | 43.5% | — |
Math gpt-oss-20b leads
Amazon Nova Pro: 28.5 (#243), gpt-oss-20b: 39.4 (#103)
| Benchmark | Amazon Nova Pro | gpt-oss-20b |
|---|---|---|
| Omni-MATH | 24.2% | 56.5% |
| LMArena Math | 1252 | 1317 |
| OTIS Mock AIME 2024-2025 | — | 65.3% |
| LiveBench Math | 38% | — |
Knowledge gpt-oss-20b leads
Amazon Nova Pro: 27.4 (#250), gpt-oss-20b: 34.6 (#195)
| Benchmark | Amazon Nova Pro | gpt-oss-20b |
|---|---|---|
| MMLU-Pro | 67.3% | 74% |
| GPQA (HELM) | 44.6% | 59.4% |
| LMArena Expert | 1211 | 1258 |
| GPQA Diamond | — | 60.8% |
| Humanity's Last Exam | 4.4% | — |
| Confabulations | 30.1% | — |
| Vectara Hallucination Rate | 5.1% | — |
| MMLU | 82% | — |
Multimodal Not comparable
Amazon Nova Pro: 25.0 (#126), gpt-oss-20b: —
| Benchmark | Amazon Nova Pro | gpt-oss-20b |
|---|---|---|
| LMArena Vision | 980 | — |
Multilingual gpt-oss-20b leads
Amazon Nova Pro: 39.7 (#223), gpt-oss-20b: 42.2 (#197)
| Benchmark | Amazon Nova Pro | gpt-oss-20b |
|---|---|---|
| LMArena Non-English | 1234 | 1268 |
| LMArena Chinese | 1244 | 1314 |
| LMArena German | 1243 | 1255 |
| LMArena Japanese | 1200 | 1244 |
| LMArena Korean | 1203 | 1236 |
| LMArena Russian | 1240 | 1278 |
| LMArena Spanish | 1182 | 1267 |
| LMArena French | 1271 | — |
Instruction Following Amazon Nova Pro leads
Amazon Nova Pro: 64.9 (#226), gpt-oss-20b: 61.8 (#240)
| Benchmark | Amazon Nova Pro | gpt-oss-20b |
|---|---|---|
| IFEval | 81.5% | 73.2% |
| LMArena Instruction Following | 1235 | 1236 |
| LiveBench Instruction Following | 67.1% | — |
Long Context Too close to call
Amazon Nova Pro: 38.1 (#205), gpt-oss-20b: 37.9 (#209)
| Benchmark | Amazon Nova Pro | gpt-oss-20b |
|---|---|---|
| LMArena Longer Query | 1255 | 1250 |
Writing & Preference Amazon Nova Pro leads
Amazon Nova Pro: 43.9 (#226), gpt-oss-20b: 35.5 (#265)
| Benchmark | Amazon Nova Pro | gpt-oss-20b |
|---|---|---|
| LMArena Text | 1259 | 1287 |
| LMArena Creative Writing | 1212 | 1201 |
| WildBench | 77.7% | 73.7% |
| LMArena Multi-Turn | 1246 | 1268 |
| Short-Story Creative Writing | 60.5% | — |
| EQ-Bench Creative Writing | — | 666 |
| LiveBench Language | 37% | — |
Frequently asked questions
Is Amazon Nova Pro better than gpt-oss-20b?
gpt-oss-20b is the stronger model overall, scoring 32.5 to 31.0 on the Noometry Index.
Which is cheaper, Amazon Nova Pro or gpt-oss-20b?
gpt-oss-20b is cheaper. It lists at $0.018 per million input tokens and $0.09 per million output tokens; Amazon Nova Pro lists at $0.80 and $3.20.
Is Amazon Nova Pro or gpt-oss-20b better for coding?
gpt-oss-20b scores higher on coding benchmarks: 37.6 versus 35.1 in the Noometry coding category.
Which has the bigger context window?
Amazon Nova Pro does, with 300K tokens against 131K.
How many benchmarks do Amazon Nova Pro and gpt-oss-20b share?
22 benchmarks have published results for both models. Amazon Nova Pro has 38 scored results on Noometry and gpt-oss-20b has 34.