Reasoning
198-question Diamond subset; independent AA evaluation
- 01
- 02
- 03
- 03
- 03
- 11
DeepSeek V4 Pro 0813
Across the task benchmarks we track, DeepSeek V4 Pro 0813’s strongest relative results are in terminal use, reasoning, and work automations. On Terminal-Bench 2.1, it scores 78.7%, ranking #4 among 13 models with reported results in our dataset.
Updated: Data as of:
Strongest results
| Benchmark | Measures | Score |
|---|---|---|
| GPQA ↗Diamond / Artificial Analysis | Reasoning | 92.83%#11 of 30 |
| AutomationBench-AA ↗1.0.6 / AA private 657 | Work automations | 56.71%#23 of 55 |
| Humanity's Last Exam ↗Stable HLE — text-only subset | Expert reasoning | 41%#30 of 68 |
| Terminal-Bench ↗4.0 / mini-swe-agent | Working in a terminal | 14.14%Joint #29 of 56 |
Model details
| Model | DeepSeek V4 Pro 0813 |
|---|---|
| Agent Index v1.2 | Insufficient coverage (4/7) · Preview |
| Creator | DeepSeek (opens in a new tab) |
| Availability | Open source / open weights |
| Context size | 1,000,000 tokens (opens in a new tab) |
| Max output | 393,216 tokens (opens in a new tab) |
| Knowledge cutoff | Not reported |
| License | MIT · Model weights ↗ (opens in a new tab) |
Pricing and performance
Per provider and reported configuration. Prices are USD per 1M tokens.
| Provider / configuration | Contexttokens | InputUSD / 1M tokens | OutputUSD / 1M tokens | LatencyTTFT · seconds | Speedtokens / sec |
|---|---|---|---|---|---|
| DeepSeek V4 Pro 0813 (max) | 1,000,000 | $1.32 (opens in a new tab) | $3.96 (opens in a new tab) | 1.589 (opens in a new tab) | 143.882 (opens in a new tab) |
| DeepSeek V4 Pro 0813 (max) | 1,000,000 | $1.32 (opens in a new tab) | $3.96 (opens in a new tab) | Not reported | Not reported |
| DeepSeek V4 Pro 0813 (max) | 1,000,000 | $1.30 (opens in a new tab) | $2.60 (opens in a new tab) | 1.129 (opens in a new tab) | 78.412 (opens in a new tab) |
| DeepSeek V4 Pro 0813 (max) | 1,048,576 | $1.32 (opens in a new tab) | $3.96 (opens in a new tab) | 1.793 (opens in a new tab) | 89.952 (opens in a new tab) |
| DeepSeek V4 Pro 0813 (max) | 1,048,576 | $1.32 (opens in a new tab) | $3.96 (opens in a new tab) | 0.832 (opens in a new tab) | 57.677 (opens in a new tab) |
| DeepSeek V4 Pro 0813 (max) | 1,048,576 | $1.056 (opens in a new tab) | $3.168 (opens in a new tab) | 3.152 (opens in a new tab) | 71.407 (opens in a new tab) |
| DeepSeek V4 Pro 0813 (max) | 1,048,576 | $1.32 (opens in a new tab) | $3.96 (opens in a new tab) | 1.602 (opens in a new tab) | 172.165 (opens in a new tab) |
| DeepSeek V4 Pro 0813 (max) | 1,048,576 | $1.32 (opens in a new tab) | $3.96 (opens in a new tab) | 2.123 (opens in a new tab) | 72.307 (opens in a new tab) |
| DeepSeek V4 Pro 0813 (max) | 1,048,576 | $1.32 (opens in a new tab) | $3.96 (opens in a new tab) | 2.088 (opens in a new tab) | 64.823 (opens in a new tab) |
| DeepSeek V4 Pro 0813 (max) | 1,048,576 | $0.50 (opens in a new tab) | $2.00 (opens in a new tab) | 1.894 (opens in a new tab) | 305.002 (opens in a new tab) |
Benchmark performance
DeepSeek V4 Pro 0813 in color, alongside leading models. Latest stored reports, best effort per report date.
GPQA Diamond ↗
Terminal Use
89 tasks; Terminus 2; e2b; three repeats
- 01
- 02
- 03
- 04
- 05
Work Automations
Private 657 tasks; 50 turns; objective completion
- 01
- 02
- 03
- 04
- 05
- 23
Knowledge Work
220 tasks; Stirrup agent; pairwise Elo
- 01
- 02
- 03
- 04
- 05
- 23
Reported benchmarks
All published results in our dataset, with versions, evaluation settings, and sources.
| Benchmark | Score | Evaluation setup | Source |
|---|---|---|---|
| AA-Briefcase1.1 / combined EloCombined Elo · Higher is better | 1,254.44 Elo | selection: best reported configuration per release · DeepSeek V4 Pro 0813 (Max) · tools: benchmark-defined | Artificial Analysis (opens in a new tab)independent |
| AutomationBench-AA1.0.6 / AA private 657Objectives completed without guardrail violations · Higher is better | 34.909% | DeepSeek V4 Pro 0813 (Non-reasoning); Private 657 tasks; 50 turns; objective completion | Artificial Analysis (opens in a new tab)independent |
| AutomationBench-AA1.0.6 / AA private 657Objectives completed without guardrail violations · Higher is better | 56.712% | DeepSeek V4 Pro 0813 (Max); Private 657 tasks; 50 turns; objective completion | Artificial Analysis (opens in a new tab)independent |
| GDP.pdf100 tasks / All-pass / five attemptsAccuracy · Higher is better | 11.4% | selection: best reported configuration per release · DeepSeek V4 Pro 0813 (Max) · tools: benchmark-defined | Artificial Analysis (opens in a new tab)independent |
| GDPval-AA2.1Elo · Higher is better | 1,066.05 Elo | DeepSeek V4 Pro 0813 (Non-reasoning); 220 tasks; Stirrup agent; pairwise Elo | Artificial Analysis (opens in a new tab)independent |
| GDPval-AA2.1Elo · Higher is better | 1,457.37 Elo | DeepSeek V4 Pro 0813 (Max); 220 tasks; Stirrup agent; pairwise Elo | Artificial Analysis (opens in a new tab)independent |
| GPQADiamond / Artificial AnalysisAccuracy · Higher is better | 92.828% | DeepSeek V4 Pro 0813 (Max); 198-question Diamond subset; independent AA evaluation | Artificial Analysis (opens in a new tab)independent |
| HLE-Diamond1,000 questions / no toolsAccuracy · Higher is better | 19.5% | selection: best reported configuration per release · DeepSeek V4 Pro 0813 (Max) · No tools | Artificial Analysis (opens in a new tab)independent |
| Humanity's Last ExamHLE — source does not specify subsetAccuracy · Higher is better | 42.7% | subset: unspecified · No tools | DeepSeek ↗ (opens in a new tab)provider reported · 2026-08-13 |
| Humanity's Last ExamHLE — source does not specify subsetAccuracy · Higher is better | 60% | subset: unspecified · Tools enabled | DeepSeek ↗ (opens in a new tab)provider reported · 2026-08-13 |
| Humanity's Last ExamStable HLE — text-only subsetAccuracy · Higher is better | 41% | question count: 2158 · reported precision decimals: 1 · DeepSeek V4 Pro 0813 (max); original HLE; 2,158 text-only questions; no tools; pass@1; chart labels rounded to one decimal. · source model href: /models/deepseek-v4-pro · subset: text · No tools | Artificial Analysis (opens in a new tab)independent |
| Terminal-Bench2.1 / Artificial AnalysisPass@1 · Higher is better | 78.652% | DeepSeek V4 Pro 0813 (Max); 89 tasks; Terminus 2; e2b; three repeats | Artificial Analysis (opens in a new tab)independent |
| Terminal-Bench4.0 / mini-swe-agentTask success · Higher is better | 14.141% | harness: mini-swe-agent · repeats: 3 · selection: best reported configuration per release · DeepSeek V4 Pro 0813 (Max) | Artificial Analysis (opens in a new tab)independent |