Model comparison
DeepSeek-V3.2-Exp vs Mistral Large 4
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 43.1 on the Noometry Index.
Last verified . 14 shared benchmarks.
DeepSeek
Mistral AISummary
- They share 14 benchmarks with published results for both. DeepSeek-V3.2-Exp scores higher in 4 categories and Mistral Large 4 in 4 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-V3.2-Exp leads 51.7 to 36.6.
- The biggest single-benchmark swing is NYT Connections (extended): 36.7% for DeepSeek-V3.2-Exp and 27.4% for Mistral Large 4.
- DeepSeek-V3.2-Exp is cheaper at $0.26 / $0.38 per million input/output tokens, against $0.68 / $2.09 for Mistral Large 4.
- Mistral Large 4 accepts more context: 1.05M tokens versus 164K.
- DeepSeek-V3.2-Exp has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3.2-Exp | Mistral Large 4 | |
|---|---|---|
| Provider | DeepSeek | Mistral AI |
| Noometry Index | 44.3 | 43.1 |
| Released | 2025-09-29 | 2026-10-06 |
| Weights | Open | Proprietary |
| Context window | 164K | 1.05M |
| Max output | 66K | 262K |
| Input $ / M tokens | $0.26 | $0.68 |
| Output $ / M tokens | $0.38 | $2.09 |
| Results tracked | 49 | 15 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Mistral Large 4 leads
DeepSeek-V3.2-Exp: 46.5 (#65), Mistral Large 4: 48.6 (#57)
| Benchmark | DeepSeek-V3.2-Exp | Mistral Large 4 |
|---|---|---|
| LMArena WebDev | 1362 | 1541 |
| LMArena Coding | 1454 | 1475 |
| SWE-bench Verified (bash only) | 70% | — |
| Aider Polyglot | 74.2% | — |
| SWE-bench Multilingual | 59% | — |
| SciCode | 38.9% | — |
| WeirdML | 39.5% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3.2-Exp: 32.7 (#59), Mistral Large 4: —
| Benchmark | DeepSeek-V3.2-Exp | Mistral Large 4 |
|---|---|---|
| Terminal-Bench | 39.6% | — |
| APEX-Agents | 21.3% | — |
| Berkeley Function Calling Leaderboard | 56.7% | — |
| TheAgentCompany | 42.9% | — |
| Vending-Bench 2 | 1,034 | — |
Reasoning Too close to call
DeepSeek-V3.2-Exp: 22.1 (#208), Mistral Large 4: 22.5 (#192)
| Benchmark | DeepSeek-V3.2-Exp | Mistral Large 4 |
|---|---|---|
| NYT Connections (extended) | 36.7% | 27.4% |
| LMArena Hard Prompts | 1434 | 1444 |
| ARC-AGI-2 | 4% | — |
| Kagi LLM Benchmark | 52.2% | — |
| ARC-AGI-1 | 57% | — |
| CritPt | 2.9% | — |
| Chess Puzzles | 14% | — |
| Thematic Generalization | 65% | — |
| DTBench | 87.7% | — |
| LMCA | 29.1% | — |
| Epoch Capabilities Index | 146.27 | — |
Math DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 41.7 (#87), Mistral Large 4: 40.4 (#91)
| Benchmark | DeepSeek-V3.2-Exp | Mistral Large 4 |
|---|---|---|
| LMArena Math | 1435 | 1488 |
| MathArena Final-Answer Competitions | 57.7% | — |
| OTIS Mock AIME 2024-2025 | 87.8% | — |
| ProofBench | 8% | — |
| FrontierMath (Feb 2025 set) | 22.1% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 51.7 (#66), Mistral Large 4: 36.6 (#166)
| Benchmark | DeepSeek-V3.2-Exp | Mistral Large 4 |
|---|---|---|
| LMArena Expert | 1436 | 1447 |
| GPQA Diamond | 83.4% | — |
| SimpleQA Verified | — | 20% |
| Vectara Hallucination Rate | 5.3% | — |
Multilingual Too close to call
DeepSeek-V3.2-Exp: 52.2 (#90), Mistral Large 4: 52.6 (#82)
| Benchmark | DeepSeek-V3.2-Exp | Mistral Large 4 |
|---|---|---|
| LMArena Non-English | 1409 | 1415 |
| LMArena Chinese | 1461 | 1491 |
| LMArena Russian | 1424 | 1414 |
| LMArena French | 1433 | — |
| LMArena German | 1440 | — |
| LMArena Japanese | 1374 | — |
| LMArena Korean | 1371 | — |
| LMArena Spanish | 1440 | — |
Instruction Following Too close to call
DeepSeek-V3.2-Exp: 74.5 (#93), Mistral Large 4: 75.0 (#76)
| Benchmark | DeepSeek-V3.2-Exp | Mistral Large 4 |
|---|---|---|
| LMArena Instruction Following | 1413 | 1424 |
Long Context DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 47.6 (#16), Mistral Large 4: 43.6 (#89)
| Benchmark | DeepSeek-V3.2-Exp | Mistral Large 4 |
|---|---|---|
| LMArena Longer Query | 1428 | 1429 |
| Fiction.LiveBench | 83.3% | — |
| CL-bench | 13.2% | — |
| CL-bench Life | 9.5% | — |
Writing & Preference DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 62.4 (#77), Mistral Large 4: 60.4 (#97)
| Benchmark | DeepSeek-V3.2-Exp | Mistral Large 4 |
|---|---|---|
| LMArena Text | 1425 | 1427 |
| LMArena Creative Writing | 1403 | 1361 |
| LMArena Multi-Turn | 1427 | 1424 |
| EQ-Bench Creative Writing | 1515 | — |
Frequently asked questions
Is DeepSeek-V3.2-Exp better than Mistral Large 4?
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 43.1 on the Noometry Index.
Which is cheaper, DeepSeek-V3.2-Exp or Mistral Large 4?
DeepSeek-V3.2-Exp is cheaper. It lists at $0.26 per million input tokens and $0.38 per million output tokens; Mistral Large 4 lists at $0.68 and $2.09.
Is DeepSeek-V3.2-Exp or Mistral Large 4 better for coding?
Mistral Large 4 scores higher on coding benchmarks: 48.6 versus 46.5 in the Noometry coding category.
Which has the bigger context window?
Mistral Large 4 does, with 1.05M tokens against 164K.
How many benchmarks do DeepSeek-V3.2-Exp and Mistral Large 4 share?
14 benchmarks have published results for both models. DeepSeek-V3.2-Exp has 49 scored results on Noometry and Mistral Large 4 has 15.