Model comparison
DeepSeek-V3.2-Exp vs o3-mini
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 36.7 on the Noometry Index.
Last verified . 32 shared benchmarks.
DeepSeekSummary
- They share 32 benchmarks with published results for both. DeepSeek-V3.2-Exp scores higher in 8 categories and o3-mini in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in long context, where DeepSeek-V3.2-Exp leads 47.6 to 33.8.
- The biggest single-benchmark swing is Fiction.LiveBench: 83.3% for DeepSeek-V3.2-Exp and 50% for o3-mini.
- DeepSeek-V3.2-Exp is cheaper at $0.26 / $0.38 per million input/output tokens, against $1.10 / $4.40 for o3-mini.
- o3-mini accepts more context: 200K tokens versus 164K.
- DeepSeek-V3.2-Exp has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3.2-Exp | o3-mini | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 44.3 | 36.7 |
| Released | 2025-09-29 | 2024-12-20 |
| Weights | Open | Proprietary |
| Context window | 164K | 200K |
| Max output | 66K | 100K |
| Input $ / M tokens | $0.26 | $1.10 |
| Output $ / M tokens | $0.38 | $4.40 |
| Results tracked | 49 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 46.5 (#65), o3-mini: 40.8 (#132)
| Benchmark | DeepSeek-V3.2-Exp | o3-mini |
|---|---|---|
| Aider Polyglot | 74.2% | 60.4% |
| SciCode | 38.9% | 39.8% |
| WeirdML | 39.5% | 43.7% |
| LMArena Coding | 1454 | 1378 |
| SWE-bench Verified (bash only) | 70% | — |
| LMArena WebDev | 1362 | — |
| SWE-bench Multilingual | 59% | — |
| GSO | — | 1.3% |
| LiveBench Coding | — | 82.7% |
| CadEval | — | 54% |
Agentic & Tool Use DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 32.7 (#59), o3-mini: 29.6 (#84)
| Benchmark | DeepSeek-V3.2-Exp | o3-mini |
|---|---|---|
| Terminal-Bench | 39.6% | — |
| APEX-Agents | 21.3% | — |
| Berkeley Function Calling Leaderboard | 56.7% | — |
| TheAgentCompany | 42.9% | — |
| Cybench | — | 22.5% |
| Vending-Bench 2 | 1,034 | — |
Reasoning DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 22.1 (#208), o3-mini: 16.3 (#305)
| Benchmark | DeepSeek-V3.2-Exp | o3-mini |
|---|---|---|
| ARC-AGI-2 | 4% | 3% |
| ARC-AGI-1 | 57% | 34.5% |
| CritPt | 2.9% | 0.3% |
| Chess Puzzles | 14% | 17% |
| LMArena Hard Prompts | 1434 | 1366 |
| DTBench | 87.7% | 68.8% |
| LMCA | 29.1% | 19% |
| Epoch Capabilities Index | 146.27 | 140.34 |
| SimpleBench | — | 22.8% |
| Kagi LLM Benchmark | 52.2% | — |
| NYT Connections (extended) | 36.7% | — |
| Thematic Generalization | 65% | — |
| LiveBench Reasoning | — | 89.6% |
| Mystery Game Puzzles | — | 7% |
| LiveBench Data Analysis | — | 70.6% |
| ForecastBench | — | 59.6 |
| LiveBench | — | 75.9% |
Math DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 41.7 (#87), o3-mini: 28.1 (#244)
| Benchmark | DeepSeek-V3.2-Exp | o3-mini |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 87.8% | 76.9% |
| LMArena Math | 1435 | 1396 |
| FrontierMath (Feb 2025 set) | 22.1% | 12.4% |
| FrontierMath Tier 4 (v1) | 2.1% | 4.2% |
| FrontierMath (Tiers 1-3) | — | 18.6% |
| FrontierMath Tier 4 | — | 0% |
| MathArena Final-Answer Competitions | 57.7% | — |
| ProofBench | 8% | — |
| LiveBench Math | — | 77.3% |
| MATH Level 5 | — | 96.5% |
Knowledge DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 51.7 (#66), o3-mini: 38.3 (#146)
| Benchmark | DeepSeek-V3.2-Exp | o3-mini |
|---|---|---|
| GPQA Diamond | 83.4% | 77% |
| LMArena Expert | 1436 | 1364 |
| SimpleQA Verified | — | 15.3% |
| Confabulations | — | 17.9% |
| Vectara Hallucination Rate | 5.3% | — |
Multilingual DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 52.2 (#90), o3-mini: 45.7 (#164)
| Benchmark | DeepSeek-V3.2-Exp | o3-mini |
|---|---|---|
| LMArena Non-English | 1409 | 1319 |
| LMArena Chinese | 1461 | 1379 |
| LMArena French | 1433 | 1334 |
| LMArena German | 1440 | 1303 |
| LMArena Japanese | 1374 | 1286 |
| LMArena Korean | 1371 | 1314 |
| LMArena Russian | 1424 | 1304 |
| LMArena Spanish | 1440 | 1321 |
Instruction Following Too close to call
DeepSeek-V3.2-Exp: 74.5 (#93), o3-mini: 75.1 (#72)
| Benchmark | DeepSeek-V3.2-Exp | o3-mini |
|---|---|---|
| LMArena Instruction Following | 1413 | 1337 |
| LiveBench Instruction Following | — | 84.4% |
Long Context DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 47.6 (#16), o3-mini: 33.8 (#256)
| Benchmark | DeepSeek-V3.2-Exp | o3-mini |
|---|---|---|
| Fiction.LiveBench | 83.3% | 50% |
| LMArena Longer Query | 1428 | 1343 |
| CL-bench | 13.2% | — |
| CL-bench Life | 9.5% | — |
Writing & Preference DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 62.4 (#77), o3-mini: 50.3 (#182)
| Benchmark | DeepSeek-V3.2-Exp | o3-mini |
|---|---|---|
| LMArena Text | 1425 | 1337 |
| LMArena Creative Writing | 1403 | 1286 |
| LMArena Multi-Turn | 1427 | 1320 |
| Short-Story Creative Writing | — | 61.7% |
| EQ-Bench Creative Writing | 1515 | — |
| LiveBench Language | — | 50.7% |
Frequently asked questions
Is DeepSeek-V3.2-Exp better than o3-mini?
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 36.7 on the Noometry Index.
Which is cheaper, DeepSeek-V3.2-Exp or o3-mini?
DeepSeek-V3.2-Exp is cheaper. It lists at $0.26 per million input tokens and $0.38 per million output tokens; o3-mini lists at $1.10 and $4.40.
Is DeepSeek-V3.2-Exp or o3-mini better for coding?
DeepSeek-V3.2-Exp scores higher on coding benchmarks: 46.5 versus 40.8 in the Noometry coding category.
Which has the bigger context window?
o3-mini does, with 200K tokens against 164K.
How many benchmarks do DeepSeek-V3.2-Exp and o3-mini share?
32 benchmarks have published results for both models. DeepSeek-V3.2-Exp has 49 scored results on Noometry and o3-mini has 51.