Model comparison
DeepSeek-V2.5 (Sep 2024) vs gpt-oss-20b
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 32.5 on the Noometry Index.
Last verified . 16 shared benchmarks.
DeepSeekSummary
- They share 16 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 6 categories and gpt-oss-20b in 2 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V2.5 (Sep 2024) leads 49.8 to 35.5.
Side by side
| DeepSeek-V2.5 (Sep 2024) | gpt-oss-20b | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 37.6 | 32.5 |
| Released | 2024-09-06 | 2025-08-05 |
| Weights | Open | Open |
| Context window | — | 131K |
| Max output | — | 16K |
| Input $ / M tokens | — | $0.018 |
| Output $ / M tokens | — | $0.09 |
| Results tracked | 22 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding gpt-oss-20b leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), gpt-oss-20b: 37.6 (#192)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | gpt-oss-20b |
|---|---|---|
| LMArena Coding | 1309 | 1306 |
| Aider Polyglot | 17.8% | — |
| SciCode | — | 34.4% |
| WeirdML | — | 40.9% |
| BigCodeBench Instruct | 48.6% | — |
| BigCodeBench Complete | 53.2% | — |
| ALE-Bench | — | 566.05 |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |
Agentic & Tool Use Not comparable
DeepSeek-V2.5 (Sep 2024): —, gpt-oss-20b: 9.3 (#154)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | gpt-oss-20b |
|---|---|---|
| Terminal-Bench | — | 3.4% |
Reasoning DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), gpt-oss-20b: 19.3 (#261)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | gpt-oss-20b |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1274 |
| Kagi LLM Benchmark | — | 53.2% |
| CritPt | — | 1.4% |
| Chess Puzzles | — | 4% |
| DTBench | — | 68% |
| LMCA | — | 14.5% |
| Epoch Capabilities Index | — | 137.82 |
Math gpt-oss-20b leads
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), gpt-oss-20b: 39.4 (#103)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | gpt-oss-20b |
|---|---|---|
| LMArena Math | 1288 | 1317 |
| OTIS Mock AIME 2024-2025 | — | 65.3% |
| Omni-MATH | — | 56.5% |
Knowledge Too close to call
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), gpt-oss-20b: 34.6 (#195)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | gpt-oss-20b |
|---|---|---|
| LMArena Expert | 1266 | 1258 |
| GPQA Diamond | — | 60.8% |
| MMLU-Pro | — | 74% |
| GPQA (HELM) | — | 59.4% |
Multilingual Too close to call
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), gpt-oss-20b: 42.2 (#197)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | gpt-oss-20b |
|---|---|---|
| LMArena Non-English | 1273 | 1268 |
| LMArena Chinese | 1318 | 1314 |
| LMArena German | 1258 | 1255 |
| LMArena Japanese | 1228 | 1244 |
| LMArena Korean | 1209 | 1236 |
| LMArena Russian | 1289 | 1278 |
| LMArena Spanish | 1248 | 1267 |
| LMArena French | 1289 | — |
Instruction Following DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), gpt-oss-20b: 61.8 (#240)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | gpt-oss-20b |
|---|---|---|
| LMArena Instruction Following | 1280 | 1236 |
| IFEval | — | 73.2% |
Long Context DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), gpt-oss-20b: 37.9 (#209)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | gpt-oss-20b |
|---|---|---|
| LMArena Longer Query | 1301 | 1250 |
Writing & Preference DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), gpt-oss-20b: 35.5 (#265)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | gpt-oss-20b |
|---|---|---|
| LMArena Text | 1294 | 1287 |
| LMArena Creative Writing | 1285 | 1201 |
| LMArena Multi-Turn | 1297 | 1268 |
| EQ-Bench Creative Writing | — | 666 |
| WildBench | — | 73.7% |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than gpt-oss-20b?
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 32.5 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or gpt-oss-20b better for coding?
gpt-oss-20b scores higher on coding benchmarks: 37.6 versus 31.7 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and gpt-oss-20b share?
16 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and gpt-oss-20b has 34.