Solar System
loads high
Tech. quality91
UX81
Time5 min 45 s
Cost$0.047
Compare →
DeepSeek V4.1 Flash
Loads without errors5/5#1 of 26
Tech. quality96/100#6 of 26
UX92/100#2 of 26
Total cost$0.24
Average time5 min 54 s#18 of 26
Total tokens408,374
API price: $0.15 / $0.6 per 1M input / output tokens · as of 16/10/2026 source ↗
What is DeepSeek V4.1 Flash good for?
The smallest model of DeepSeek's new architecture, with native visual understanding, aimed at fast inference and high throughput.
Description based on the maker · source ↗
- Maker
- DeepSeek
- Released
- 10/09/2026
- Reasoning levels
- low high max
How did DeepSeek V4.1 Flash do in the AI Lab tests?
- 5 of 5 outputs load without errors.
- Its best technical result was on the “GTA-style game” prompt (100/100).
- It implemented the chess rules flawlessly – passing the objective perft test on all 6 positions.
- Average time per prompt 5 min 54 s#18 of 26
- Average cost per prompt $0.049#8 of 26
- Technical quality 96/100#6 of 26
- Automatic UX rating 92/100#2 of 26
Outputs
All outputs of DeepSeek V4.1 Flash on the 5 prompts every other model got as well.
GTA-style game
loads high
Tech. quality100
UX100
Time5 min 24 s
Cost$0.044
Compare →
Monitor test
loads high
Tech. quality92
UX92
Time4 min 33 s
Cost$0.037
Compare →
New York & Central Park map
loads high
Tech. quality97
UX87
Time6 min 57 s
Cost$0.056
Compare →
Chess with an engine
loads high
Tech. quality100
UX99
Time6 min 53 s
Cost$0.060
Compare →
Cost = tokens priced with the API price list valid on the date shown next to the cost (prices change). ≈ = estimate (only the total token count is known), no ≈ = exact from input / output tokens. “Loads” = the page opens without errors; technical quality also checks behaviour during use – judge the quality in the viewer.