Model comparison
DeepSeek-R1 vs gpt-oss-20b
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 32.5 on the Noometry Index. gpt-oss-20b costs 25× less per token, which makes it the better buy when DeepSeek-R1's lead doesn't matter for your workload.
Last verified . 30 shared benchmarks.
DeepSeekSummary
- They share 30 benchmarks with published results for both. DeepSeek-R1 scores higher in 8 categories and gpt-oss-20b in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-R1 leads 61.4 to 35.5.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 69.4% for DeepSeek-R1 and 53.2% for gpt-oss-20b.
- gpt-oss-20b is cheaper at $0.018 / $0.09 per million input/output tokens, against $0.50 / $2.15 for DeepSeek-R1.
- DeepSeek-R1 accepts more context: 164K tokens versus 131K.
- gpt-oss-20b has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-R1 | gpt-oss-20b | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 42.3 | 32.5 |
| Released | 2025-01-20 | 2025-08-05 |
| Weights | Proprietary | Open |
| Context window | 164K | 131K |
| Max output | 64K | 16K |
| Input $ / M tokens | $0.50 | $0.018 |
| Output $ / M tokens | $2.15 | $0.09 |
| Results tracked | 52 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 46.3 (#68), gpt-oss-20b: 37.6 (#192)
| Benchmark | DeepSeek-R1 | gpt-oss-20b |
|---|---|---|
| SciCode | 35.7% | 34.4% |
| WeirdML | 41.6% | 40.9% |
| LMArena Coding | 1427 | 1306 |
| ALE-Bench | 804.12 | 566.05 |
| Aider Polyglot | 71.4% | — |
| LiveBench Coding | 66.7% | — |
| AlgoTune | 1.7 | — |
Agentic & Tool Use DeepSeek-R1 leads
DeepSeek-R1: 30.7 (#75), gpt-oss-20b: 9.3 (#154)
| Benchmark | DeepSeek-R1 | gpt-oss-20b |
|---|---|---|
| Terminal-Bench | — | 3.4% |
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
| METR Time Horizons | 53.8% | — |
Reasoning Too close to call
DeepSeek-R1: 18.6 (#278), gpt-oss-20b: 19.3 (#261)
| Benchmark | DeepSeek-R1 | gpt-oss-20b |
|---|---|---|
| Kagi LLM Benchmark | 69.4% | 53.2% |
| CritPt | 1.1% | 1.4% |
| LMArena Hard Prompts | 1416 | 1274 |
| Epoch Capabilities Index | 141.29 | 137.82 |
| ARC-AGI-2 | 1.3% | — |
| SimpleBench | 40.8% | — |
| ARC-AGI-1 | 21.2% | — |
| Chess Puzzles | — | 4% |
| LiveBench Reasoning | 83.2% | — |
| DTBench | — | 68% |
| LiveBench Data Analysis | 69.8% | — |
| LMCA | — | 14.5% |
| ForecastBench | 60 | — |
| LiveBench | 71.6% | — |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), gpt-oss-20b: 39.4 (#103)
| Benchmark | DeepSeek-R1 | gpt-oss-20b |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 65.3% |
| Omni-MATH | 42.4% | 56.5% |
| LMArena Math | 1400 | 1317 |
| LiveBench Math | 80.7% | — |
| MATH Level 5 | 96.6% | — |
Knowledge DeepSeek-R1 leads
DeepSeek-R1: 44.5 (#87), gpt-oss-20b: 34.6 (#195)
| Benchmark | DeepSeek-R1 | gpt-oss-20b |
|---|---|---|
| GPQA Diamond | 76.3% | 60.8% |
| MMLU-Pro | 79.3% | 74% |
| GPQA (HELM) | 66.6% | 59.4% |
| LMArena Expert | 1394 | 1258 |
| Confabulations | 12.7% | — |
| Vectara Hallucination Rate | 11.3% | — |
Multilingual DeepSeek-R1 leads
DeepSeek-R1: 52.4 (#85), gpt-oss-20b: 42.2 (#197)
| Benchmark | DeepSeek-R1 | gpt-oss-20b |
|---|---|---|
| LMArena Non-English | 1412 | 1268 |
| LMArena Chinese | 1442 | 1314 |
| LMArena German | 1404 | 1255 |
| LMArena Japanese | 1391 | 1244 |
| LMArena Korean | 1360 | 1236 |
| LMArena Russian | 1423 | 1278 |
| LMArena Spanish | 1411 | 1267 |
| LMArena French | 1417 | — |
Instruction Following DeepSeek-R1 leads
DeepSeek-R1: 72.0 (#143), gpt-oss-20b: 61.8 (#240)
| Benchmark | DeepSeek-R1 | gpt-oss-20b |
|---|---|---|
| IFEval | 78.4% | 73.2% |
| LMArena Instruction Following | 1382 | 1236 |
| LiveBench Instruction Following | 80.5% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), gpt-oss-20b: 37.9 (#209)
| Benchmark | DeepSeek-R1 | gpt-oss-20b |
|---|---|---|
| LMArena Longer Query | 1391 | 1250 |
| Fiction.LiveBench | 75% | — |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.4 (#88), gpt-oss-20b: 35.5 (#265)
| Benchmark | DeepSeek-R1 | gpt-oss-20b |
|---|---|---|
| LMArena Text | 1428 | 1287 |
| LMArena Creative Writing | 1405 | 1201 |
| EQ-Bench Creative Writing | 1500 | 666 |
| WildBench | 82.8% | 73.7% |
| LMArena Multi-Turn | 1405 | 1268 |
| Short-Story Creative Writing | 83% | — |
| LiveBench Language | 48.5% | — |
Frequently asked questions
Is DeepSeek-R1 better than gpt-oss-20b?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 32.5 on the Noometry Index. gpt-oss-20b costs 25× less per token, which makes it the better buy when DeepSeek-R1's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-R1 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; DeepSeek-R1 lists at $0.50 and $2.15.
Is DeepSeek-R1 or gpt-oss-20b better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 37.6 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-R1 does, with 164K tokens against 131K.
How many benchmarks do DeepSeek-R1 and gpt-oss-20b share?
30 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and gpt-oss-20b has 34.