Noometry

Model comparison

DeepSeek V4 Flash vs Gemini 2.5 Pro

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 45.0 on the Noometry Index.

Last verified . 35 shared benchmarks.

DeepSeek V4 Flash DeepSeek

53.6

Rank #35 Confirmed

Gemini 2.5 Pro Google

45.0

Rank #75 Confirmed

Summary

  • They share 35 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 4 categories and Gemini 2.5 Pro in 4 categories; 5 gaps are clear of the uncertainty.
  • The widest gap is in math, where DeepSeek V4 Flash leads 60.3 to 32.5.
  • The biggest single-benchmark swing is ARC-AGI-2: 61.4% for DeepSeek V4 Flash and 4.9% for Gemini 2.5 Pro.
  • DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $1.25 / $10 for Gemini 2.5 Pro.
  • Gemini 2.5 Pro accepts more context: 1.05M tokens versus 1M.
  • DeepSeek V4 Flash has downloadable open weights; the other is API-only.

Side by side

DeepSeek V4 FlashGemini 2.5 Pro
ProviderDeepSeekGoogle
Noometry Index53.645.0
Released2026-04-242025-03-25
WeightsOpenProprietary
Context window1M1.05M
Max output393K66K
Input $ / M tokens$0.15$1.25
Output $ / M tokens$0.60$10
Results tracked4178

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

Category by category

Coding DeepSeek V4 Flash leads

DeepSeek V4 Flash: 47.9 (#59), Gemini 2.5 Pro: 42.4 (#101)

BenchmarkDeepSeek V4 FlashGemini 2.5 Pro
LMArena WebDev15821227
SciCode49.9%42.8%
WeirdML63%54%
LMArena Coding14571452
ALE-Bench1,306785.52
SWE-bench Verified—57.6%
FrontierCode18.8%—
SWE-bench Verified (bash only)—53.6%
Aider Polyglot—83.1%
GSO—3.9%
LiveBench Coding—85.9%
CadEval—64%
AlgoTune—1.51

Agentic & Tool Use Not comparable

DeepSeek V4 Flash: —, Gemini 2.5 Pro: 29.2 (#88)

BenchmarkDeepSeek V4 FlashGemini 2.5 Pro
Terminal-Bench—32.6%
GDPval—23.3%
Remote Labor Index—0.8%
TheAgentCompany—30.3%
τ²-bench Banking—13.7%
DeepResearch Bench—42.8%
BALROG—43.3%
LMArena Search—1142
METR Time Horizons—55.4%
Vending-Bench 2—573.64

Reasoning DeepSeek V4 Flash leads

DeepSeek V4 Flash: 53.7 (#30), Gemini 2.5 Pro: 28.8 (#99)

BenchmarkDeepSeek V4 FlashGemini 2.5 Pro
ARC-AGI-261.4%4.9%
SimpleBench61.1%62.4%
Kagi LLM Benchmark52.2%70.3%
ARC-AGI-189%41%
CritPt16.6%2%
Chess Puzzles33%20%
LMArena Hard Prompts14441455
DTBench90.9%82.4%
LMCA41.7%34.8%
Epoch Capabilities Index154.49145.32
NYT Connections (extended)89.6%—
EnigmaEval—5.6%
LiveBench Reasoning—89.8%
Mystery Game Puzzles34%—
LiveBench Data Analysis—79.9%
ForecastBench—61.3
LiveBench—82.3%

Math DeepSeek V4 Flash leads

DeepSeek V4 Flash: 60.3 (#37), Gemini 2.5 Pro: 32.5 (#213)

BenchmarkDeepSeek V4 FlashGemini 2.5 Pro
FrontierMath (Tiers 1-3)57.5%24.6%
FrontierMath Tier 424.4%0%
OTIS Mock AIME 2024-202594.4%84.7%
LMArena Math14271450
MathArena Final-Answer Competitions76.5%—
ProofBench56%—
Omni-MATH—41.6%
LiveBench Math—90.2%
MATH Level 5—95.9%
FrontierMath (Feb 2025 set)—14.1%
FrontierMath Tier 4 (v1)—4.2%

Knowledge Too close to call

DeepSeek V4 Flash: 55.4 (#48), Gemini 2.5 Pro: 56.0 (#46)

BenchmarkDeepSeek V4 FlashGemini 2.5 Pro
GPQA Diamond91%85.3%
LMArena Expert14411452
Humanity's Last Exam—21.6%
SimpleQA Verified33.6%—
MMLU-Pro—86.3%
Confabulations—10.6%
Vectara Hallucination Rate—7%
GPQA (HELM)—74.9%

Multimodal Not comparable

DeepSeek V4 Flash: —, Gemini 2.5 Pro: 45.2 (#18)

BenchmarkDeepSeek V4 FlashGemini 2.5 Pro
LMArena Vision—1263
GeoBench—86%
VPCT—48%
LMArena Document—1421
SpatialViz-Bench—44.7%

Multilingual Gemini 2.5 Pro leads

DeepSeek V4 Flash: 53.0 (#72), Gemini 2.5 Pro: 55.3 (#31)

BenchmarkDeepSeek V4 FlashGemini 2.5 Pro
LMArena Non-English14201451
LMArena Chinese14681507
LMArena French14391472
LMArena German14181487
LMArena Japanese14061461
LMArena Korean13841434
LMArena Russian14281461
LMArena Spanish14361473

Instruction Following Too close to call

DeepSeek V4 Flash: 74.9 (#81), Gemini 2.5 Pro: 75.0 (#75)

BenchmarkDeepSeek V4 FlashGemini 2.5 Pro
LMArena Instruction Following14211437
LiveBench Instruction Following—80.6%
IFEval—84%

Long Context Gemini 2.5 Pro leads

DeepSeek V4 Flash: 43.8 (#85), Gemini 2.5 Pro: 59.8 (#5)

BenchmarkDeepSeek V4 FlashGemini 2.5 Pro
LMArena Longer Query14341449
Fiction.LiveBench—91.7%

Writing & Preference Too close to call

DeepSeek V4 Flash: 63.8 (#61), Gemini 2.5 Pro: 63.7 (#62)

BenchmarkDeepSeek V4 FlashGemini 2.5 Pro
LMArena Text14321458
LMArena Creative Writing14031454
EQ-Bench Creative Writing15591421
LMArena Multi-Turn14491453
Short-Story Creative Writing—83.8%
WildBench—85.7%
LiveBench Language—67.8%

Frequently asked questions

Is DeepSeek V4 Flash better than Gemini 2.5 Pro?

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 45.0 on the Noometry Index.

Which is cheaper, DeepSeek V4 Flash or Gemini 2.5 Pro?

DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; Gemini 2.5 Pro lists at $1.25 and $10.

Is DeepSeek V4 Flash or Gemini 2.5 Pro better for coding?

DeepSeek V4 Flash scores higher on coding benchmarks: 47.9 versus 42.4 in the Noometry coding category.

Which has the bigger context window?

Gemini 2.5 Pro does, with 1.05M tokens against 1M.

How many benchmarks do DeepSeek V4 Flash and Gemini 2.5 Pro share?

35 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and Gemini 2.5 Pro has 78.

Related comparisons

Go deeper