- API price from
- ≈ $0.05 in / ≈ $0.14 out / 1M tokOpenRouter · DeepInfra · FP8 · promotional
- Context window
- 1.05M tok
Capabilities
Context window
Modalities
Integration
Pricing & access
APIUSD per 1M tokens
- Input
- $0.30
- Output
- $1.20
- Cached input
- $0.006
Run locally
- Original download
- 510.3 GB
- App memory
- Unverified
- Documented software
- Author inference codeSetup guide (opens in a new tab)
Find GGUF downloadsBenchmarks
Source (opens in a new tab)Reported by DeepSeek
12 benchmarks · Select a result for detailsAgents’ Last Exam35.4 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 35.4 points
reported score · deepseek-1017-author-report
Headline result. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)
Apex Bench v1.172.1 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 72.1 points
reported score · deepseek-1017-author-report
Headline result. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)
Automation-Bench v1.0.660.5 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 60.5 points
reported score · deepseek-1017-author-report
Headline result. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)
BabyVision (w/tools)92.5 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 92.5 points
reported score · deepseek-1017-author-report
Headline result. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)
Chartography (w/tools)83.0 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 83.0 points
reported score · deepseek-1017-author-report
Headline result. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)
DeepSWE v1.177.0 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 77.0 points
reported score · deepseek-1017-author-report
Headline result. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)
HLE w/o tools44.0 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 44.0 points
reported score · deepseek-1017-author-report
Text-only subset. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)
HLE w/o tools42.6 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 42.6 points
reported score · deepseek-1017-author-report
Headline result. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)
NL2Repo-Bench68.2 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 68.2 points
reported score · deepseek-1017-author-report
Headline result. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)
ProgramBench29.1 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 29.1 points
reported score · deepseek-1017-author-report
Headline result. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)
Terminal-Bench 4.042.3 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 42.3 points
reported score · deepseek-1017-author-report
Headline result. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)
ZeroBench-main (w/tools)51.0 points
Reported by DeepSeek · DeepSeek-V4.1-Flash-1017
Reported value: 51.0 points
reported score · deepseek-1017-author-report
Headline result. Author-reported score; unreported effort, sample size and evaluation date remain unknown. Separate from independent benchmark cohorts.
Protocol incomplete; no ranking assigned.
Sample size: Not reported · Standard error: Not reported
Captured 2026-10-17 · Source (opens in a new tab)