Reasoning
No current rating
Mathematical and scientific problem solving.
Awaiting a qualified category score
Model profile / DeepSeek
DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows.
Four areas of evidence
Mathematical and scientific problem solving.
Software engineering and programming.
Factual reliability and instruction following.
Tool use, workflows and professional tasks.
Current index snapshot · 2026-09-16. Each category uses its own scale; scores across categories are not directly comparable. Counts describe published catalogue evidence. A point lead does not establish superiority.
Performance in context
DeepSeek V4 Flash 0731 has no current Capabilities rating. Explore the rated models below.
Lumina Capabilities Index
The current Capabilities cohort, with available operational measurements.
Chart loads as you explore
Lumina Capabilities Index
The current Capabilities cohort, with available operational measurements.
Chart loads as you explore
Explore the detail
No current index estimate. Available benchmark measurements and source records are retained below.
| Benchmark | Result | Published configuration | Source |
|---|---|---|---|
| Agents Last Exam2026 | 25.2% | DeepSeek-V4-Flash-0731 official model card; exact ALE configuration not specifiedSource-native system | DeepSeek-V4-Flash-0731 official model card and provider evaluation ↗Observed 2026-07-31 · checked 2026-08-24provider-reported · reference-only |
| CyberGymstandard | 76.7% | DeepSeek Harness Minimal Mode; max effort; top_p=0.95; temperature=1.0DeepSeek Harness Minimal Mode | DeepSeek-V4-Flash-0731 official model card and provider evaluation ↗Observed 2026-07-31 · checked 2026-08-24provider-reported · reference-only |
| DeepSWEv1.1 | 54.4% | DeepSeek Harness Minimal Mode; max effort; top_p=0.95; temperature=1.0DeepSeek Harness Minimal Mode | DeepSeek-V4-Flash-0731 official model card and provider evaluation ↗Observed 2026-07-31 · checked 2026-08-24provider-reported · reference-only |
| Terminal-Bench2.1 | 82.7% | DeepSeek Harness Minimal Mode; max effort; top_p=0.95; temperature=1.0DeepSeek Harness Minimal Mode | DeepSeek-V4-Flash-0731 official model card and provider evaluation ↗Observed 2026-07-31 · checked 2026-08-24provider-reported · reference-only |
| Agents Last ExamAugust 2026 | 25.2% | DeepSeek V4 Flash 0731DeepSeek V4 Pro 0813 official GA comparison table | DeepSeek permanent refresh source ↗Observed 2026-08-13 · checked 2026-08-13provider-reported · reference-only |
| AutomationBenchPublic 600-task set, August 2026 | 25.1% | DeepSeek V4 Flash 0731AutomationBench public set; DeepSeek V4 Pro 0813 official GA comparison table | DeepSeek permanent refresh source ↗Observed 2026-08-13 · checked 2026-08-13provider-reported · reference-only |
| CyberGymrolling | 76.7% | DeepSeek V4 Flash 0731DeepSeek V4 Pro 0813 official GA comparison table | DeepSeek permanent refresh source ↗Observed 2026-08-13 · checked 2026-08-13provider-reported · reference-only |
| DeepSWEv1.1 | 54.4% | DeepSeek V4 Flash 0731DeepSeek V4 Pro 0813 official GA comparison table | DeepSeek permanent refresh source ↗Observed 2026-08-13 · checked 2026-08-13provider-reported · reference-only |
| DeepSeek DSBench FullStackAugust 2026 | 68.7% | DeepSeek V4 Flash 0731DeepSeek V4 Pro 0813 official GA comparison table | DeepSeek permanent refresh source ↗Observed 2026-08-13 · checked 2026-08-13provider-reported · reference-only |
| DeepSeek DSBench HardAugust 2026 | 59.6% | DeepSeek V4 Flash 0731DeepSeek V4 Pro 0813 official GA comparison table | DeepSeek permanent refresh source ↗Observed 2026-08-13 · checked 2026-08-13provider-reported · reference-only |