Leaderboard
Language ModelsOpen weights
DeepSeek V3.2
Open-weights MoE (671B total / 37B active), MIT; predecessor to V4.
Crosshair Index
61.2
#25 of 32 · Language Models
- Provider
- DeepSeek
- Released
- 2025-12-01
- Parameters
- 671B
- Context
- 128K tokens
Token pricing
- Input
- $0.28 /1M
- Output
- $0.42 /1M
- Cache read
- $0.18 /1M
USD per 1M tokens · cache read = cached input (hit), cache write = caching surcharge · official list pricing (June 2026).
Industry skill web
Professional-domain strengths, composed from the benchmarks relevant to each field. Highlight an axis to see the benchmarks behind it.
Medicine
84
skill
Clinical knowledge and diagnostic reasoning, including the medical coding accuracy and science depth that real practice demands.
Scorecard
| Benchmark | Score | Source | Status |
|---|---|---|---|
| GPQA Diamond Reasoning | 82.4% | DeepSeek — V3.2 technical reportvendor | unverified |
| Humanity's Last Exam Frontier | 25.1% | DeepSeek — V3.2 technical reportvendor | unverified |
| SWE-bench Verified Agentic Coding | 73.1% | DeepSeek — V3.2 technical reportvendor | unverified |
| LiveCodeBench Coding | 83.3% | DeepSeek — V3.2 technical reportvendor | unverified |
| MMLU-Pro Knowledge | 85% | DeepSeek — V3.2 technical reportvendor | unverified |
| LMArena Elo Human Preference | 1,437 | LMArena (arena.ai)3rd-party | unverified |
| AA Intelligence Index Composite | 32 | Artificial Analysis3rd-party | unverified |
| Corporate Finance Finance | 51% | Vals AI — CorpFin v23rd-party | unverified |
| LegalBench Law | 76.1% | Vals AI — LegalBench3rd-party | unverified |
| TaxEval Tax & Accounting | 68.2% | Vals AI — TaxEval v23rd-party | unverified |
| Medical Coding Medicine | — | not evaluated |