DeepSeek V4 Pro 0813 (max): benchmark scores, price and ranking
DeepSeek V4 Pro 0813 (max) from DeepSeek ranks #25 of 741 LLMs on the LLMs Tiger aggregate benchmark score (54.3), based on 12 benchmark results captured up to 2026-10-05.
- Organisation
- DeepSeek
- Type
- Open-weight / local model
- Open weights
- Yes
- Released
- 2026-08-13 (source: epoch)
- Cheapest API price
- $0.660 in / $1.98 out per million tokens (via openrouter, observed 2026-10-05T21:47Z)
- Aggregate rank
- #25 of 741 (score 54.3)
Relative to other models, DeepSeek V4 Pro 0813 (max) is strongest on Chess Puzzles (#10 of 124) and weakest on FrontierMath T4 (#31 of 52).
DeepSeek V4 Pro 0813 (max) benchmark results
| Benchmark | Score | Rank | Source | Captured |
|---|---|---|---|---|
| Chess Puzzles | 47.0% | #10 of 124 | Epoch AI Benchmarking Hub | 2026-10-05 |
| OTIS Mock AIME | 98.6% | #16 of 188 | Epoch AI Benchmarking Hub | 2026-10-05 |
| GPQA Diamond | 91.7% | #24 of 274 | Epoch AI Benchmarking Hub | 2026-10-05 |
| ALE-Bench | 1,403 | #10 of 78 | Epoch AI Benchmarking Hub | 2026-10-05 |
| WeirdML | 66.2% | #16 of 112 | Epoch AI Benchmarking Hub | 2026-10-05 |
| CritPt | 18.0% | #23 of 111 | Epoch AI Benchmarking Hub | 2026-10-05 |
| Mystery Games | 43.0% | #12 of 57 | Epoch AI Benchmarking Hub | 2026-10-05 |
| SimpleQA Verified | 52.9% | #14 of 66 | Epoch AI Benchmarking Hub | 2026-10-05 |
| SciCode | 51.0% | #29 of 100 | Epoch AI Benchmarking Hub | 2026-10-05 |
| ARC-AGI-2 | 61.2% | #28 of 89 | ARC Prize official leaderboard | 2026-10-05 |
| FrontierMath T1-3 | 64.6% | #29 of 69 | Epoch AI Benchmarking Hub | 2026-10-05 |
| FrontierMath T4 | 26.8% | #31 of 52 | Epoch AI Benchmarking Hub | 2026-10-05 |
DeepSeek V4 Pro 0813 (max) is also listed at 5 other reasoning-effort settings; this page shows the highest-effort row, which is the one the leaderboard keeps by default.
Models ranked nearby
- #22 gemini-4-argon-high (54.5)
- #23 Claude 4.5 Opus (54.5)
- #24 Claude 3.5-Sonnet-20241022 (54.4)
- #26 GPT-5.3 Codex (xhigh) (54.2)
- #27 Qwen 3.6 Max (Preview) (54.1)
- #28 GPT-6 Astra (pro, max) (54.0)
Compare it on the live leaderboard → · How the aggregate score works