DeepSeek V4 Flash 0731
By DeepSeek · released 2026-07-31 (2 months) · 1 effort level
1 of 1 on the open-weights frontier at launchOff the frontier todayopen weights
- Intelligence index
- 34.30
- Cost to run the AA index
- $200
- Price per 1M tokens (in / out)
- $0.14 / $0.28
Today
Where DeepSeek V4 Flash 0731 stands on the 2026-10-25 frontier
No measured level is on the frontier today.
| Level | Input modalities | Intelligence | Cost | E2E | Today | Matchup |
|---|---|---|---|---|---|---|
| DeepSeek V4 Flash 0731 (Max) | Text | 34.30 | $200 | 17.6s | beaten by Claude Haiku 5.5 (Xhigh) | Compare |
DeepSeek V4 Flash 0731 effort levelsToday's Pareto frontierOpen-weights frontierEvery other model
Matchup
Compare head to head
Suggested: what beats it, its effort levels, and frontier models near its price.
- DeepSeek V4 Flash 0731 (Max) vs Claude Haiku 5.5 (Xhigh)
- DeepSeek V4 Flash 0731 (Max) vs MiMo-V2.6-Pro
- DeepSeek V4 Flash 0731 (Max) vs GPT-6 Luna (Xhigh)
- DeepSeek V4 Flash 0731 (Max) vs Claude Haiku 5.5 (High)
- DeepSeek V4 Flash 0731 (Max) vs MiMo-V2.6-Flash
- DeepSeek V4 Flash 0731 (Max) vs GPT-6 Luna (Max)
- DeepSeek V4 Flash 0731 (Max) vs GPT-6.1 Sol (Medium)
- DeepSeek V4 Flash 0731 (Max) vs GPT-6.1 Sol (High)
Launch day
What DeepSeek V4 Flash 0731 did to the frontier on 2026-07-31
Reached the open-weights frontier
DeepSeek V4 Flash 0731 (Max)
No level beat the overall frontier, but no earlier open-weights model was as smart for less.
Reached the frontier0 of 1
Impact (hypervolume gained)0
Rank, past year's releases—
Smartest level34.30
Cheapest level$200
Reached the frontierMissed itDisplaced from the old frontierFrontier the day beforeFrontier after the releaseRegion gained (impact)
Missed the frontier
| Level | Intelligence | Cost | Beaten by | Intel gap | Cost ratio |
|---|---|---|---|---|---|
| DeepSeek V4 Flash 0731 (Max) | 34.30 | $200 | GPT-5.6 Luna (Xhigh) | −0.3 | 1.11× |
Agents:
Accept: text/markdown or ?format=md returns this page as Markdown. Markdown: /releases/deepseek-v4-flash-0731?format=md · llms.txt · agent skill