Compare Models

DeepSeek V4 Pro 0813

DeepSeek V4 Pro 0813 is an open-weight model from DeepSeek, released August 13, 2026. It ranks #27 of 36 models on the Vals Index with 52.97%. Its best result is #27 of 127 on Vibe Code Bench v1.1.

Release Date: Aug 13, 2026

Developer DeepSeek  🇨🇳
Context Window 1M
Max Output Tokens 384k
Token Costs (in/out) $1.32/3.96
Weights Open
Input Modalities

Accuracy

52.97 % ± 1.04

Cost / Test (Vals Index)

$ 2.500

Latency

58 min 3 s

Vals Index
BenchmarksAccuracyRankings
Vals Index

52.97%

±1.04
27/36
Vals Finance Index

52.94%

±1.41
33/39
Vals Legal Index

69.01%

±1.77
21/37
Code Migration v1.1

51.99%

±2.23
15/30
CyberBench v1.1

72.86%

±5.50
14/48
EMB v1.1

65.13%

±2.35
34/41
Finance Agent v2.1

57.12%

±2.08
31/39
Immigration

61.66%

±1.68
24/26
Legal Research Bench

40.87%

±3.42
35/93
MedCode

42.47%

±2.16
50/107
MedScribe

80.17%

±2.00
59/109
ProofBench v1.1

50.00%

±5.03
32/53
SRE Bench

1.91%

±0.85
26/37
Tax Agent Bench

26.76%

±3.09
41/84
TaxEval v2

73.06%

±0.87
55/145
Vibe Code Bench 1-100

17.53%

±3.41
16/26
Vibe Code Bench v1.1

82.30%

±3.15
27/127
HLAB-Vals-1.1

10.83%

±2.72
21/37
GPQA Diamond

92.42%

±2.02
17/138
IOI

51.61%

±2.40
26/46
LiveCodeBench

87.53%

±0.96
13/143
LegalBench

82.36%

±0.44
73/151
MMLU Pro

86.97%

±0.34
36/138
ProgramBench

0.00%

±0.00
35/66
SkillsBench

53.83%

±4.49
16/35
SWE-bench

96.40%

±0.83
2/88
Terminal-Bench 4.0

14.14%

±2.02
41/63
Terminal-Bench-Science

4.29%

±2.44
32/45
Proprietary BenchmarksAcademic BenchmarksIndustry Partners
Vals
Default Provider : DeepSeek
Temperature: 1
Top P: Default
Top K: Default
Max Output Tokens: 384,000
Reasoning Effort: max

Compared with

DeepSeek V4 Pro 0813 vs Qwen 3.8 Max 21 shared benchmarks DeepSeek V4 Pro 0813 vs Muse Spark Open 11 shared benchmarks DeepSeek V4 Pro 0813 vs Mistral Large 4 Preview 15 shared benchmarks DeepSeek V4 Pro 0813 vs Muse Spark 1.2 17 shared benchmarks DeepSeek V4 Pro 0813 vs Gemini 3.5 Flash 16 shared benchmarks DeepSeek V4 Pro 0813 vs Grok 4.5 18 shared benchmarks DeepSeek V4 Pro 0813 vs GLM 5.3 Flash 26 shared benchmarks DeepSeek V4 Pro 0813 vs Step 5 Preview 15 shared benchmarks

Updates

Aug 12, 2026

We evaluated DeepSeek’s DeepSeek V4 Pro 0813 across our benchmark suite.

  • It ranks #18 on the Vals Index (52.37%), up 9.48 points from DeepSeek V4 (42.89%).

  • Its standout result is SWE-bench Verified, where it places #2 of 82 models (96.40%) and is the highest-scoring open-weight model on the board, ahead of Kimi K3 (93.40%). It is also by far the cheapest model near the top: $0.02 per test, versus $1.29 for Claude Opus 5 at 97.00%.

  • Reasoning and legal results improve sharply over the previous release: 49.00% on ProofBench (up from 10.00%, #45 → #15) and 40.87% on Legal Research Bench (up from 23.08%, #27 → #11). It reaches 7.50% on Harvey’s Legal Agent Benchmark with an 88.06% criteria pass rate, #10 of 43.

  • It struggles on terminal-driven coding and Excel tasks: it scores 54.68% across three full trials of Terminal-Bench 2.1 (#33 of 52, 28.89% on hard tasks) and 52.80% on EMB (#24 of 37).

The model has a 1M-token context window and supports up to 384k output tokens and tool calling. Evaluations were run at max reasoning effort; the model does not accept a temperature parameter.

Congrats to the DeepSeek team on the release!