DeepSeek V4.1 Flash vs Gemini 3.7 Flash: Benchmark Comparison

DeepSeek V4.1 Flash has the higher score on 8 of 14 shared benchmarks; Gemini 3.7 Flash leads on 6.

The largest observed score gap is 29.94 pts on CyberBench v1.1 , where DeepSeek V4.1 Flash leads.

Reported ±1 standard-error ranges overlap on 5 of 14 shared benchmarks with comparable uncertainty data. This is not a pairwise statistical significance test.

Shared benchmark results

Scores and reported standard errors come from the latest visible Vals benchmark version available for both models.

Benchmark DeepSeek V4.1 Flash Gemini 3.7 Flash Gap Reported uncertainty
Legal Research Bench 41.35% ±3.42 34.62% ±3.31 6.73 pts Reported ±1 SE ranges do not overlap
LegalBench 83.28% ±0.46 87.26% ±0.42 3.98 pts Reported ±1 SE ranges do not overlap
Tax Agent Bench 26.63% ±3.14 22.80% ±3.03 3.83 pts Reported ±1 SE ranges overlap
MedCode 41.17% ±2.04 53.39% ±2.12 12.22 pts Reported ±1 SE ranges do not overlap
MedScribe 85.50% ±1.92 83.94% ±2.00 1.56 pts Reported ±1 SE ranges overlap
ProofBench v1.1 54.00% ±5.01 58.00% ±4.96 4.00 pts Reported ±1 SE ranges overlap
Terminal-Bench-Science 0.00% ±0.00 7.14% ±3.10 7.14 pts Reported ±1 SE ranges do not overlap
IOI 40.28% ±2.56 67.83% ±3.61 27.55 pts Reported ±1 SE ranges do not overlap
ProgramBench 0.50% ±0.50 0.00% ±0.00 0.50 pts Reported ±1 SE ranges overlap
SkillsBench 69.80% ±3.88 65.89% ±4.55 3.91 pts Reported ±1 SE ranges overlap
Terminal-Bench 4.0 19.70% ±1.75 12.12% ±3.03 7.58 pts Reported ±1 SE ranges do not overlap
Vibe Code Bench v1.1 84.74% ±2.89 70.39% ±4.84 14.34 pts Reported ±1 SE ranges do not overlap
CyberBench v1.1 73.69% ±5.48 43.75% ±2.21 29.94 pts Reported ±1 SE ranges do not overlap
SRE Bench 0.76% ±0.54 4.58% ±1.29 3.82 pts Reported ±1 SE ranges do not overlap

Performance by category

Category DeepSeek V4.1 Flash average Gemini 3.7 Flash average
Legal 62.31% 60.94%
Finance 26.63% 22.80%
Healthcare 63.34% 68.67%
Math 54.00% 58.00%
Science 0.00% 7.14%
Coding 43.00% 43.25%
Cyber 37.23% 24.16%

Cost and latency

Cost per test appears only where the benchmark reports it for both models. Latency is the measured completion time for that benchmark.

Benchmark DeepSeek V4.1 Flash cost Gemini 3.7 Flash cost DeepSeek V4.1 Flash latency Gemini 3.7 Flash latency
Legal Research Bench $0.25 $0.76 11m42s 4m49s
LegalBench N/A N/A 6.53s 2.23s
Tax Agent Bench $0.13 $0.48 5m49s 87.24s
MedCode N/A N/A 35.13s 9.16s
MedScribe N/A N/A 53.83s 16.82s
ProofBench v1.1 $0.13 $0.56 9m07s 7m04s
Terminal-Bench-Science $0.47 $7.19 2h57m 1h20m
IOI $0.24 $3.40 18m30s 21m09s
ProgramBench $0.90 $7.13 1h02m 21m30s
SkillsBench $0.09 $1.80 3m43s 3m06s
Terminal-Bench 4.0 $0.50 $8.71 48m22s 1h24m
Vibe Code Bench v1.1 $0.41 $4.83 15m28s 23m43s
CyberBench v1.1 $0.11 $2.15 13m50s 7m34s
SRE Bench $0.55 $21.47 1h01m 1h50m

Results available only for DeepSeek V4.1 Flash

  • Vals Finance Index
  • Vals Science Index
  • Vals Inference Provider Index
  • Immigration
  • EMB v1.1
  • Finance Agent v2.1
  • BioMysteryBench
  • MysteryMechanism
  • SAGE
  • Code Migration v1.1
  • SWE-bench Multimodal
  • Access to Justice Benchmark
  • Financial Coaching Bench
  • Public Benefits Bench v1.1

Results available only for Gemini 3.7 Flash

  • Vals RSI Index
  • MortgageTax
  • TaxEval v2
  • GPQA Diamond
  • MMLU Pro
  • MMMU Pro
  • LiveCodeBench
  • SWE-bench
Model details DeepSeek V4.1 Flash Model details Gemini 3.7 Flash