Model comparison

DeepSeek-V3.2-Exp vs Gemini 2.5 Flash

DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 39.3 on the Noometry Index.

Last verified October 10, 2026. 37 shared benchmarks.

DeepSeek-V3.2-Exp

DeepSeek

44.3

Rank #78 Confirmed

Gemini 2.5 Flash

Google

39.3

Rank #170 Confirmed

Summary

They share 37 benchmarks with published results for both. DeepSeek-V3.2-Exp scores higher in 7 categories and Gemini 2.5 Flash in 2 categories; 7 gaps are clear of the uncertainty.
The widest gap is in knowledge, where DeepSeek-V3.2-Exp leads 51.7 to 36.4.
The biggest single-benchmark swing is SWE-bench Verified (bash only): 70% for DeepSeek-V3.2-Exp and 28.7% for Gemini 2.5 Flash.
DeepSeek-V3.2-Exp is cheaper at $0.26 / $0.38 per million input/output tokens, against $0.30 / $2.50 for Gemini 2.5 Flash.
Gemini 2.5 Flash accepts more context: 1.05M tokens versus 164K.
DeepSeek-V3.2-Exp has downloadable open weights; the other is API-only.

Side by side

DeepSeek-V3.2-Exp and Gemini 2.5 Flash specifications
	DeepSeek-V3.2-Exp	Gemini 2.5 Flash
Provider	DeepSeek	Google
Noometry Index	44.3	39.3
Released	2025-09-29	2025-04-17
Weights	Open	Proprietary
Context window	164K	1.05M
Max output	66K	66K
Input $ / M tokens	$0.26	$0.30
Output $ / M tokens	$0.38	$2.50
Results tracked	49	54

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

Category by category

Coding DeepSeek-V3.2-Exp leads

DeepSeek-V3.2-Exp: 46.5 (#65), Gemini 2.5 Flash: 35.8 (#220)

Coding benchmarks
Benchmark	DeepSeek-V3.2-Exp	Gemini 2.5 Flash
SWE-bench Verified (bash only)	70%	28.7%
Aider Polyglot	74.2%	55.1%
WeirdML	39.5%	41.9%
LMArena Coding	1454	1424
LMArena WebDev	1362	—
SWE-bench Multilingual	59%	—
SciCode	38.9%	—
ALE-Bench	—	661.88

Agentic & Tool Use DeepSeek-V3.2-Exp leads

DeepSeek-V3.2-Exp: 32.7 (#59), Gemini 2.5 Flash: 30.8 (#74)

Agentic & Tool Use benchmarks
Benchmark	DeepSeek-V3.2-Exp	Gemini 2.5 Flash
Terminal-Bench	39.6%	17.1%
Berkeley Function Calling Leaderboard	56.7%	56.2%
TheAgentCompany	42.9%	41.1%
Vending-Bench 2	1,034	548.84
APEX-Agents	21.3%	—
BALROG	—	33.5%

Reasoning DeepSeek-V3.2-Exp leads

DeepSeek-V3.2-Exp: 22.1 (#208), Gemini 2.5 Flash: 18.1 (#286)

Reasoning benchmarks
Benchmark	DeepSeek-V3.2-Exp	Gemini 2.5 Flash
ARC-AGI-2	4%	2.5%
Kagi LLM Benchmark	52.2%	56.8%
ARC-AGI-1	57%	33.3%
CritPt	2.9%	1.1%
LMArena Hard Prompts	1434	1422
DTBench	87.7%	76.5%
LMCA	29.1%	27.5%
Epoch Capabilities Index	146.27	143.03
SimpleBench	—	41.2%
NYT Connections (extended)	36.7%	—
Chess Puzzles	14%	—
EnigmaEval	—	2.7%
Thematic Generalization	65%	—
ForecastBench	—	60.6

Math DeepSeek-V3.2-Exp leads

DeepSeek-V3.2-Exp: 41.7 (#87), Gemini 2.5 Flash: 39.9 (#98)

Math benchmarks
Benchmark	DeepSeek-V3.2-Exp	Gemini 2.5 Flash
OTIS Mock AIME 2024-2025	87.8%	73.1%
LMArena Math	1435	1415
FrontierMath (Feb 2025 set)	22.1%	4.8%
FrontierMath Tier 4 (v1)	2.1%	4.2%
MathArena Final-Answer Competitions	57.7%	—
ProofBench	8%	—
Omni-MATH	—	38.5%

Knowledge DeepSeek-V3.2-Exp leads

DeepSeek-V3.2-Exp: 51.7 (#66), Gemini 2.5 Flash: 36.4 (#168)

Knowledge benchmarks
Benchmark	DeepSeek-V3.2-Exp	Gemini 2.5 Flash
Vectara Hallucination Rate	5.3%	7.8%
LMArena Expert	1436	1426
GPQA Diamond	83.4%	—
Humanity's Last Exam	—	12.1%
MMLU-Pro	—	63.9%
Confabulations	—	16.8%
GPQA (HELM)	—	39%

Multimodal Not comparable

DeepSeek-V3.2-Exp: —, Gemini 2.5 Flash: 41.8 (#32)

Multimodal benchmarks
Benchmark	DeepSeek-V3.2-Exp	Gemini 2.5 Flash
LMArena Vision	—	1253
GeoBench	—	76%
VPCT	—	46.2%
SpatialViz-Bench	—	36.9%

Multilingual Too close to call

DeepSeek-V3.2-Exp: 52.2 (#90), Gemini 2.5 Flash: 52.3 (#88)

Multilingual benchmarks
Benchmark	DeepSeek-V3.2-Exp	Gemini 2.5 Flash
LMArena Non-English	1409	1409
LMArena Chinese	1461	1450
LMArena French	1433	1433
LMArena German	1440	1418
LMArena Japanese	1374	1405
LMArena Korean	1371	1385
LMArena Russian	1424	1415
LMArena Spanish	1440	1421

Instruction Following Gemini 2.5 Flash leads

DeepSeek-V3.2-Exp: 74.5 (#93), Gemini 2.5 Flash: 75.7 (#54)

Instruction Following benchmarks
Benchmark	DeepSeek-V3.2-Exp	Gemini 2.5 Flash
LMArena Instruction Following	1413	1405
IFEval	—	89.8%

Long Context Too close to call

DeepSeek-V3.2-Exp: 47.6 (#16), Gemini 2.5 Flash: 47.5 (#17)

Long Context benchmarks
Benchmark	DeepSeek-V3.2-Exp	Gemini 2.5 Flash
Fiction.LiveBench	83.3%	77.8%
LMArena Longer Query	1428	1419
CL-bench	13.2%	—
CL-bench Life	9.5%	—

Writing & Preference DeepSeek-V3.2-Exp leads

DeepSeek-V3.2-Exp: 62.4 (#77), Gemini 2.5 Flash: 53.8 (#157)

Writing & Preference benchmarks
Benchmark	DeepSeek-V3.2-Exp	Gemini 2.5 Flash
LMArena Text	1425	1417
LMArena Creative Writing	1403	1400
EQ-Bench Creative Writing	1515	1137
LMArena Multi-Turn	1427	1408
Short-Story Creative Writing	—	76.5%
WildBench	—	81.7%

Frequently asked questions

Is DeepSeek-V3.2-Exp better than Gemini 2.5 Flash?

DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 39.3 on the Noometry Index.

Which is cheaper, DeepSeek-V3.2-Exp or Gemini 2.5 Flash?

DeepSeek-V3.2-Exp is cheaper. It lists at $0.26 per million input tokens and $0.38 per million output tokens; Gemini 2.5 Flash lists at $0.30 and $2.50.

Is DeepSeek-V3.2-Exp or Gemini 2.5 Flash better for coding?

DeepSeek-V3.2-Exp scores higher on coding benchmarks: 46.5 versus 35.8 in the Noometry coding category.

Which has the bigger context window?

Gemini 2.5 Flash does, with 1.05M tokens against 164K.

How many benchmarks do DeepSeek-V3.2-Exp and Gemini 2.5 Flash share?

37 benchmarks have published results for both models. DeepSeek-V3.2-Exp has 49 scored results on Noometry and Gemini 2.5 Flash has 54.