Model comparison
DeepSeek-V3.2-Exp vs GPT-4 Turbo
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.9 to 30.3 on the Noometry Index.
Last verified . 24 shared benchmarks.
DeepSeekSummary
- They share 24 benchmarks with published results for both. DeepSeek-V3.2-Exp scores higher in 8 categories and GPT-4 Turbo in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-V3.2-Exp leads 40.8 to 8.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 87.8% for DeepSeek-V3.2-Exp and 6.7% for GPT-4 Turbo.
- DeepSeek-V3.2-Exp is cheaper at $0.26 / $0.38 per million input/output tokens, against $10 / $30 for GPT-4 Turbo.
- DeepSeek-V3.2-Exp accepts more context: 164K tokens versus 128K.
- DeepSeek-V3.2-Exp has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3.2-Exp | GPT-4 Turbo | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 44.9 | 30.3 |
| Released | 2025-09-29 | 2023-11-06 |
| Weights | Open | Proprietary |
| Context window | 164K | 128K |
| Max output | 66K | 4K |
| Input $ / M tokens | $0.26 | $10 |
| Output $ / M tokens | $0.38 | $30 |
| Results tracked | 48 | 36 |
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: 46.4 (#61), GPT-4 Turbo: 33.4 (#247)
| Benchmark | DeepSeek-V3.2-Exp | GPT-4 Turbo |
|---|---|---|
| WeirdML | 39.5% | 18% |
| LMArena Coding | 1454 | 1268 |
| SWE-bench Verified (bash only) | 70% | — |
| Aider Polyglot | 74.2% | — |
| LMArena WebDev | 1362 | — |
| SWE-bench Multilingual | 59% | — |
| SciCode | 38.9% | — |
| BigCodeBench Instruct | — | 48.2% |
| BigCodeBench Complete | — | 58.2% |
| HumanEval+ | — | 86.6% |
| MBPP+ | — | 73.3% |
Agentic & Tool Use Not comparable
DeepSeek-V3.2-Exp: 37.9 (#31), GPT-4 Turbo: —
| Benchmark | DeepSeek-V3.2-Exp | GPT-4 Turbo |
|---|---|---|
| Terminal-Bench | 39.6% | — |
| Berkeley Function Calling Leaderboard | 56.7% | — |
| TheAgentCompany | 42.9% | — |
| METR Time Horizons | — | 36.7% |
| Vending-Bench 2 | 1,034 | — |
Reasoning DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 22.0 (#185), GPT-4 Turbo: 14.9 (#311)
| Benchmark | DeepSeek-V3.2-Exp | GPT-4 Turbo |
|---|---|---|
| Chess Puzzles | 14% | 6% |
| LMArena Hard Prompts | 1433 | 1251 |
| DTBench | 87.7% | 61.6% |
| LMCA | 29.1% | 9.8% |
| Epoch Capabilities Index | 146.28 | 127.25 |
| ARC-AGI-2 | 4% | — |
| SimpleBench | — | 25.1% |
| Kagi LLM Benchmark | 52.2% | — |
| NYT Connections (extended) | 36.7% | — |
| ARC-AGI-1 | 57% | — |
| CritPt | 2.9% | — |
| Thematic Generalization | 65% | — |
| ForecastBench | — | 59.4 |
Math DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 40.8 (#75), GPT-4 Turbo: 8.6 (#316)
| Benchmark | DeepSeek-V3.2-Exp | GPT-4 Turbo |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 87.8% | 6.7% |
| LMArena Math | 1435 | 1272 |
| FrontierMath (Tiers 1-3) | — | 0.7% |
| MathArena Final-Answer Competitions | 57.7% | — |
| ProofBench | 8% | — |
| MATH Level 5 | — | 46.7% |
| FrontierMath (Feb 2025 set) | 22.1% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 51.6 (#61), GPT-4 Turbo: 24.3 (#263)
| Benchmark | DeepSeek-V3.2-Exp | GPT-4 Turbo |
|---|---|---|
| GPQA Diamond | 83.4% | 46.6% |
| LMArena Expert | 1436 | 1223 |
| Confabulations | — | 28.4% |
| Vectara Hallucination Rate | 5.3% | — |
| MMLU | — | 81.3% |
Multimodal Not comparable
DeepSeek-V3.2-Exp: —, GPT-4 Turbo: 30.0 (#104)
| Benchmark | DeepSeek-V3.2-Exp | GPT-4 Turbo |
|---|---|---|
| LMArena Vision | — | 1090 |
Multilingual DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 52.4 (#85), GPT-4 Turbo: 40.7 (#211)
| Benchmark | DeepSeek-V3.2-Exp | GPT-4 Turbo |
|---|---|---|
| LMArena Non-English | 1409 | 1245 |
| LMArena Chinese | 1461 | 1242 |
| LMArena French | 1435 | 1275 |
| LMArena German | 1440 | 1259 |
| LMArena Japanese | 1377 | 1194 |
| LMArena Korean | 1372 | 1187 |
| LMArena Russian | 1423 | 1259 |
| LMArena Spanish | 1441 | 1260 |
Instruction Following DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 74.7 (#89), GPT-4 Turbo: 66.0 (#211)
| Benchmark | DeepSeek-V3.2-Exp | GPT-4 Turbo |
|---|---|---|
| LMArena Instruction Following | 1413 | 1250 |
Long Context DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 47.6 (#15), GPT-4 Turbo: 38.0 (#200)
| Benchmark | DeepSeek-V3.2-Exp | GPT-4 Turbo |
|---|---|---|
| LMArena Longer Query | 1428 | 1254 |
| Fiction.LiveBench | 83.3% | — |
| CL-bench | 13.2% | — |
| CL-bench Life | 9.5% | — |
Writing & Preference DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 62.5 (#73), GPT-4 Turbo: 47.9 (#201)
| Benchmark | DeepSeek-V3.2-Exp | GPT-4 Turbo |
|---|---|---|
| LMArena Text | 1425 | 1272 |
| LMArena Creative Writing | 1403 | 1269 |
| LMArena Multi-Turn | 1427 | 1267 |
| EQ-Bench Creative Writing | 1515 | — |
Frequently asked questions
Is DeepSeek-V3.2-Exp better than GPT-4 Turbo?
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.9 to 30.3 on the Noometry Index.
Which is cheaper, DeepSeek-V3.2-Exp or GPT-4 Turbo?
DeepSeek-V3.2-Exp is cheaper. It lists at $0.26 per million input tokens and $0.38 per million output tokens; GPT-4 Turbo lists at $10 and $30.
Is DeepSeek-V3.2-Exp or GPT-4 Turbo better for coding?
DeepSeek-V3.2-Exp scores higher on coding benchmarks: 46.4 versus 33.4 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3.2-Exp does, with 164K tokens against 128K.
How many benchmarks do DeepSeek-V3.2-Exp and GPT-4 Turbo share?
24 benchmarks have published results for both models. DeepSeek-V3.2-Exp has 48 scored results on Noometry and GPT-4 Turbo has 36.