DeepSeek-V3.2 benchmark ranking on every AI leaderboard
open weights · top 10 on 1 of 8 · released Dec 1, 2025
| Rank | Board | Score | |||
|---|---|---|---|---|---|
| 12 |  BFCLAgents and tools · as “DeepSeek-V3.2-Exp (Prompt + Thinking)” | Agents and tools | 56.7% | 81 | |
| 55 |  ARC-AGI-2Reasoning · as “Deepseek V3.2” | Reasoning | 4.0% | 78 | |
| 69 |  Epoch ECIReasoning · as “DeepSeek-V3.2” | Reasoning | 146.3 | 100 | |
| 77 |  WebDev ArenaCoding · as “deepseek-v3.2-thinking” | Coding | 1,361 | 100 | |
| 90 | Text ArenaChat · as “deepseek-v3.2-exp-thinking” | Chat | 1,425 | 100 | |
| 8 |  Aider PolyglotCoding · stale since 2025-10-03 · as “DeepSeek-V3.2-Exp (Reasoner)” | Coding | 74.2% | 50 | |
| 11 |  SWE-bench VerifiedCoding · stale since 2026-02-26 · as “DeepSeek V3.2 (high)” | Coding | 70.0% | 39 | |
| 33 |  MMLU-ProReasoning · stale since 2026-03-11 · as “DeepSeek-V3.2-Thinking” | Reasoning | 85.0% | 100 |
Category ranks and more models
Sources: each leaderboard's own published results page, captured every day at 04:30 UTC. Scores exactly as each board published them; we run no evaluations. Logos via logo.dev; trademarks belong to their owners.