Noometry

Model comparison

DeepSeek V4 Pro vs Gemini 3 Flash Preview

DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 52.3 on the Noometry Index.

Last verified . 39 shared benchmarks.

DeepSeek V4 Pro DeepSeek

54.3

Rank #31 Confirmed

Gemini 3 Flash Preview Google

52.3

Rank #40 Confirmed

Summary

  • They share 39 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 6 categories and Gemini 3 Flash Preview in 3 categories; 5 gaps are clear of the uncertainty.
  • The widest gap is in math, where DeepSeek V4 Pro leads 64.8 to 51.7.
  • The biggest single-benchmark swing is ProofBench: 50% for DeepSeek V4 Pro and 15% for Gemini 3 Flash Preview.
  • DeepSeek V4 Pro is cheaper at $0.66 / $1.98 per million input/output tokens, against $0.50 / $3 for Gemini 3 Flash Preview.
  • Gemini 3 Flash Preview accepts more context: 1.05M tokens versus 1M.
  • DeepSeek V4 Pro has downloadable open weights; the other is API-only.

Side by side

DeepSeek V4 ProGemini 3 Flash Preview
ProviderDeepSeekGoogle
Noometry Index54.352.3
Released2026-04-242025-12-17
WeightsOpenProprietary
Context window1M1.05M
Max output393K66K
Input $ / M tokens$0.66$0.50
Output $ / M tokens$1.98$3
Results tracked4859

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding DeepSeek V4 Pro leads

DeepSeek V4 Pro: 52.4 (#34), Gemini 3 Flash Preview: 50.9 (#42)

BenchmarkDeepSeek V4 ProGemini 3 Flash Preview
SWE-bench Verified77.6%75.4%
LMArena WebDev15821439
WeirdML66.2%61.6%
LMArena Coding14701460
ALE-Bench1,4031,367
FrontierCode28.6%—
SWE-bench Verified (bash only)—75.8%
SWE-bench Multilingual—72.7%
SciCode51%—
GSO—9.8%

Agentic & Tool Use Gemini 3 Flash Preview leads

DeepSeek V4 Pro: 32.8 (#58), Gemini 3 Flash Preview: 38.7 (#29)

BenchmarkDeepSeek V4 ProGemini 3 Flash Preview
Vending-Bench 23,2853,635
Terminal-Bench—64.3%
APEX-Agents47.3%—
τ²-bench Airline—82.5%
τ²-bench Banking—27.3%
τ²-bench Retail—76.8%
τ²-bench Telecom—91.2%
DeepResearch Bench—49.8%
BALROG—48.1%
GDP.pdf—10%
LMArena Search—1198

Reasoning DeepSeek V4 Pro leads

DeepSeek V4 Pro: 56.5 (#24), Gemini 3 Flash Preview: 49.2 (#37)

BenchmarkDeepSeek V4 ProGemini 3 Flash Preview
ARC-AGI-261.3%33.6%
NYT Connections (extended)91.3%83.1%
ARC-AGI-190.5%84.7%
Chess Puzzles47%40%
LMArena Hard Prompts14611465
Mystery Game Puzzles43%26%
DTBench93.9%89.1%
LMCA45.5%43.1%
Epoch Capabilities Index155.31151.8
ForecastBench56.158.5
SimpleBench—61.1%
Kagi LLM Benchmark53.5%—
CritPt18%—
Surface Evolver Bench40%—

Math DeepSeek V4 Pro leads

DeepSeek V4 Pro: 64.8 (#30), Gemini 3 Flash Preview: 51.7 (#55)

BenchmarkDeepSeek V4 ProGemini 3 Flash Preview
FrontierMath (Tiers 1-3)64.6%51.2%
FrontierMath Tier 426.8%17.1%
MathArena Final-Answer Competitions76.6%67.6%
OTIS Mock AIME 2024-202598.6%95.6%
ProofBench50%15%
LMArena Math14551473
FrontierMath (Feb 2025 set)—35.6%
FrontierMath Tier 4 (v1)—4.2%

Knowledge Too close to call

DeepSeek V4 Pro: 59.5 (#31), Gemini 3 Flash Preview: 58.8 (#33)

BenchmarkDeepSeek V4 ProGemini 3 Flash Preview
GPQA Diamond91.7%89.4%
SimpleQA Verified52.9%66.8%
Vectara Hallucination Rate8.6%13.5%
LMArena Expert14641462

Multimodal Not comparable

DeepSeek V4 Pro: —, Gemini 3 Flash Preview: 45.5 (#16)

BenchmarkDeepSeek V4 ProGemini 3 Flash Preview
LMArena Vision—1285
GeoBench—88%
VPCT—72.6%
Blueprint-Bench 2—0%
LMArena Document—1413

Multilingual Gemini 3 Flash Preview leads

DeepSeek V4 Pro: 54.4 (#45), Gemini 3 Flash Preview: 55.7 (#27)

BenchmarkDeepSeek V4 ProGemini 3 Flash Preview
LMArena Non-English14391458
LMArena Chinese14861511
LMArena French14721477
LMArena German14581497
LMArena Japanese14451489
LMArena Korean14471443
LMArena Russian14531480
LMArena Spanish14581469

Instruction Following Too close to call

DeepSeek V4 Pro: 76.1 (#47), Gemini 3 Flash Preview: 75.7 (#56)

BenchmarkDeepSeek V4 ProGemini 3 Flash Preview
LMArena Instruction Following14481437

Long Context Too close to call

DeepSeek V4 Pro: 45.0 (#51), Gemini 3 Flash Preview: 44.4 (#67)

BenchmarkDeepSeek V4 ProGemini 3 Flash Preview
LMArena Longer Query14581452
CL-bench Life13.5%—

Writing & Preference Too close to call

DeepSeek V4 Pro: 65.5 (#46), Gemini 3 Flash Preview: 65.5 (#45)

BenchmarkDeepSeek V4 ProGemini 3 Flash Preview
LMArena Text14511466
LMArena Creative Writing14461457
LMArena Multi-Turn14671471
EQ-Bench Creative Writing1553—
EQ-Bench 41166—

Frequently asked questions

Is DeepSeek V4 Pro better than Gemini 3 Flash Preview?

DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 52.3 on the Noometry Index.

Which is cheaper, DeepSeek V4 Pro or Gemini 3 Flash Preview?

DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; Gemini 3 Flash Preview lists at $0.50 and $3.

Is DeepSeek V4 Pro or Gemini 3 Flash Preview better for coding?

DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 50.9 in the Noometry coding category.

Which has the bigger context window?

Gemini 3 Flash Preview does, with 1.05M tokens against 1M.

How many benchmarks do DeepSeek V4 Pro and Gemini 3 Flash Preview share?

39 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Gemini 3 Flash Preview has 59.

Related comparisons

Go deeper