Model detail
DeepSeek V4 Flash
Provider: DeepSeek · Public slug: deepseek-deepseek-v4-flash
- Rank
- 29
- Rank movement
- Baseline
- Score
- 23.5
- Last measured
- 2026-07-05
- Coverage
- Complete coverage
Reasoning scores
- XHigh
- 28.0
- Medium
- 21.0
- None
- 21.6
Generation health
XHigh33%40%27%n 15Medium63%25%12%n 16None58%21%21%n 24 Complete coverageHigh failure rateCost summary
- Mean standard list-price equivalent
- $0.0052
- Median unique-cell cost
- $0.0030
- Unique-cell range
- $0.0018–$0.02
- Coverage
- 24 unique cells / 55 observed runs
- Observed-run total
- $0.25
- Evidence basis
- Provider-reported
- Mean output
- 13k tok
- Output range
- 1.7k–82k tok
- Token samples
- 55
Cost means equal-weight unique model/game/reasoning cells. Provider-account billed spend is reported separately on Charts.
Run setup
- Overall Models: combines public reasoning variants
- Standard GameBench code-generation prompt