Enter opens the highlighted result. Esc closes.

DeepSeek V3.2 Exp vs Gemini 2.5 Pro

DeepSeek V3.2 Exp leads Gemini 2.5 Pro on the Artificial Analysis Intelligence Index, 16.6 to 16.1, and DeepSeek V3.2 Exp costs 24.4x less per output token ($0.41 vs $10.00 per million). Updated October 22, 2026; 52 benchmarks have a result for both.

Replace DeepSeek V3.2 Exp Replace Gemini 2.5 Pro
DeepSeek V3.2 ExpDeepSeek
Coding
No data
Overall
No data
Output $/M
$0.41
12
Gemini 2.5 ProGoogle
Coding
19.1
Overall
41.8
Output $/M
$10.00
38
vs boards won of 52 shared, 2 tied

DeepSeek V3.2 Exp or Gemini 2.5 Pro: which to pick

Only Gemini 2.5 Pro has a coding score, 19.1 on the Vibe Coding Index; DeepSeek V3.2 Exp has none, so the two cannot be ranked on coding.

Pick DeepSeek V3.2 Exp for output tokens 24.4x cheaper ($0.41 vs $10.00 per million) and work measured by LMArena WebDev or Terminal-Bench Hard (Artificial Analysis run), where it leads.

Model ADeepSeek V3.2 ExpDeepSeekNot benchmarked1 independent benchmarkNo cost per solved taskModel BGemini 2.5 ProGoogle19.1/1003 independent benchmarks$0.64 per solved task
Shared results
52
DeepSeek V3.2 Exp only
7
Gemini 2.5 Pro only
38
Like-for-like categories
8 of 9

Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.

Which one for your work

Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.

At a glance

Coding index
DeepSeek V3.2 Exp not benchmarked, Gemini 2.5 Pro 19.1 from 3 benchmarks.
Widest category
Agentic work, 20.2 points apart on average over 1 shared benchmark.
Widest single gap
LMArena Text: Japanese: DeepSeek V3.2 Exp 1331 against Gemini 2.5 Pro 1461, 21.6 points apart on a 0 to 100 scale.
Output price
DeepSeek V3.2 Exp $0.41 against Gemini 2.5 Pro $10.00 per million output tokens, 24.4x apart.
Context window
DeepSeek V3.2 Exp takes 164K tokens, Gemini 2.5 Pro 1.05M.

Benchmark by benchmark

First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.

Boards won, where both have a result
CategoryDeepSeek V3.2 ExpBarsGemini 2.5 ProTies
Coding34
Intelligence11
Agentic work01
Reasoning06
Math022 ties
Knowledge05
Web design51
General318
All shared boards12382 ties
CodingLike-for-likeSix of these boards feed the Vibe Coding Index34
Coding benchmarks for DeepSeek V3.2 Exp and Gemini 2.5 Pro
BenchmarkDeepSeek V3.2 ExpGemini 2.5 ProLeader
Aider polyglot74.2%Not measuredOnly DeepSeek V3.2 Exp measured
SWE-bench VerifiedNot measured53.6%Only Gemini 2.5 Pro measured
LMArena WebDev12721226DeepSeek V3.2 Exp
AA Coding IndexNot measured33.3Only Gemini 2.5 Pro measured
SWE-bench Verified (Vals AI)Not measured54.4%Only Gemini 2.5 Pro measured
Vibe Code Bench (Vals AI)Not measured0.4%Only Gemini 2.5 Pro measured
LiveCodeBench (Artificial Analysis run)78.9%80.1%Gemini 2.5 Pro
SciCodeNot measured46.3%Only Gemini 2.5 Pro measured
Terminal-Bench Hard (Artificial Analysis run)31.1%26.5%DeepSeek V3.2 Exp
Terminal-BenchNot measured32.6%Only Gemini 2.5 Pro measured
GSO-BenchNot measured3.9%Only Gemini 2.5 Pro measured
WeirdML39.5%54.0%Gemini 2.5 Pro
LMArena Text: coding14381452Gemini 2.5 Pro
LMArena Text: software and IT14381459Gemini 2.5 Pro
LMArena WebDev: HTML12811235DeepSeek V3.2 Exp
LMArena WebDev: from an imageNot measured1285Only Gemini 2.5 Pro measured
Neither model has a result on the other 23 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index11
Intelligence benchmarks for DeepSeek V3.2 Exp and Gemini 2.5 Pro
BenchmarkDeepSeek V3.2 ExpGemini 2.5 ProLeader
AA Intelligence Index16.616.1DeepSeek V3.2 Exp
Epoch Capabilities Index144.9145.1Gemini 2.5 Pro
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
Agentic work benchmarks for DeepSeek V3.2 Exp and Gemini 2.5 Pro
BenchmarkDeepSeek V3.2 ExpGemini 2.5 ProLeader
AA Agentic IndexNot measured1.6Only Gemini 2.5 Pro measured
τ²-Bench Telecom33.9%54.1%Gemini 2.5 Pro
GDPvalNot measured23.3%Only Gemini 2.5 Pro measured
DeepResearch BenchNot measured41.5%Only Gemini 2.5 Pro measured
TheAgentCompany42.9%Not measuredOnly DeepSeek V3.2 Exp measured
Vending-Bench 2Not measured$574Only Gemini 2.5 Pro measured
Berkeley Function Calling Leaderboard (BFCL)56.7%Not measuredOnly DeepSeek V3.2 Exp measured
BrowseComp-PlusNot measured29.5%Only Gemini 2.5 Pro measured
Neither model has a result on the other 33 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index06
Reasoning benchmarks for DeepSeek V3.2 Exp and Gemini 2.5 Pro
BenchmarkDeepSeek V3.2 ExpGemini 2.5 ProLeader
GPQA Diamond (Artificial Analysis run)79.7%84.4%Gemini 2.5 Pro
Humanity's Last Exam (Artificial Analysis run)14.9%22.5%Gemini 2.5 Pro
GPQA Diamond (Epoch AI run)Not measured85.3%Only Gemini 2.5 Pro measured
ARC-AGINot measured41.0%Only Gemini 2.5 Pro measured
ARC-AGI-2Not measured4.9%Only Gemini 2.5 Pro measured
LMArena Text: hard prompts14291455Gemini 2.5 Pro
LMArena Text: expert14221452Gemini 2.5 Pro
LMArena Text: hard prompts in English14441453Gemini 2.5 Pro
LMArena Vision: diagramsNot measured1274Only Gemini 2.5 Pro measured
LMArena Vision: homeworkNot measured1281Only Gemini 2.5 Pro measured
Kagi LLM Benchmark52.2%70.3%Gemini 2.5 Pro
Neither model has a result on the other 11 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for DeepSeek V3.2 Exp and Gemini 2.5 Pro
BenchmarkDeepSeek V3.2 ExpGemini 2.5 ProLeader
AA Math Index87.787.7Tie
AIME 2025 (Artificial Analysis run)87.7%87.7%Tie
OTIS Mock AIMENot measured84.7%Only Gemini 2.5 Pro measured
FrontierMath, tiers 1 to 3Not measured24.6%Only Gemini 2.5 Pro measured
FrontierMath, tier 4Not measured0.0%Only Gemini 2.5 Pro measured
LMArena Text: math14221450Gemini 2.5 Pro
LMArena Text: mathematical work14331454Gemini 2.5 Pro
Neither model has a result on the other 6 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index05
Knowledge benchmarks for DeepSeek V3.2 Exp and Gemini 2.5 Pro
BenchmarkDeepSeek V3.2 ExpGemini 2.5 ProLeader
CaseLaw (Vals AI)Not measured63.9%Only Gemini 2.5 Pro measured
CorpFin (Vals AI)Not measured60.8%Only Gemini 2.5 Pro measured
MortgageTax (Vals AI)Not measured68.9%Only Gemini 2.5 Pro measured
MedCode (Vals AI)Not measured50.6%Only Gemini 2.5 Pro measured
MMLU Pro (Artificial Analysis run)85.0%86.2%Gemini 2.5 Pro
LMArena Text: science14431480Gemini 2.5 Pro
LMArena Text: medicine and healthcare14681476Gemini 2.5 Pro
LMArena Text: legal and government14191480Gemini 2.5 Pro
LMArena Text: business and finance14121439Gemini 2.5 Pro
FACTS Benchmark SuiteNot measured61.5%Only Gemini 2.5 Pro measured
Neither model has a result on the other 8 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index51
Web design benchmarks for DeepSeek V3.2 Exp and Gemini 2.5 Pro
BenchmarkDeepSeek V3.2 ExpGemini 2.5 ProLeader
Design Arena: websites11781167DeepSeek V3.2 Exp
Design Arena: UI components11641137DeepSeek V3.2 Exp
Design Arena: game dev11441111DeepSeek V3.2 Exp
Design Arena: data viz11521222Gemini 2.5 Pro
Design Arena: 3D11481080DeepSeek V3.2 Exp
Design Arena: all code categories11701152DeepSeek V3.2 Exp
Design Arena: SVG1031Not measuredOnly DeepSeek V3.2 Exp measured
Neither model has a result on the other 2 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index318
General benchmarks for DeepSeek V3.2 Exp and Gemini 2.5 Pro
BenchmarkDeepSeek V3.2 ExpGemini 2.5 ProLeader
SAGE (Vals AI)Not measured41.9%Only Gemini 2.5 Pro measured
MedScribe (Vals AI)Not measured73.5%Only Gemini 2.5 Pro measured
IFBench54.1%48.7%DeepSeek V3.2 Exp
Long context reasoning (AA-LCR)72.3%69.0%DeepSeek V3.2 Exp
DTBench87.7%82.4%DeepSeek V3.2 Exp
LMCA29.1%34.8%Gemini 2.5 Pro
Fiction.LiveBench83.3%Not measuredOnly DeepSeek V3.2 Exp measured
Creative writing (Lech Mazur)Not measured83.8%Only Gemini 2.5 Pro measured
CL-bench13.2%Not measuredOnly DeepSeek V3.2 Exp measured
CL-bench Life9.5%Not measuredOnly DeepSeek V3.2 Exp measured
LMArena Text14251458Gemini 2.5 Pro
LMArena Text: instruction following14011437Gemini 2.5 Pro
LMArena Text: creative writing14041454Gemini 2.5 Pro
LMArena VisionNot measured1263Only Gemini 2.5 Pro measured
LMArena Text: English14411458Gemini 2.5 Pro
LMArena Text: non-English14041452Gemini 2.5 Pro
LMArena Text: Chinese14611507Gemini 2.5 Pro
LMArena Text: German14401487Gemini 2.5 Pro
LMArena Text: Japanese13311461Gemini 2.5 Pro
LMArena Text: Korean13721434Gemini 2.5 Pro
LMArena Text: French14341473Gemini 2.5 Pro
LMArena Text: Spanish14411474Gemini 2.5 Pro
LMArena Text: Russian14121461Gemini 2.5 Pro
LMArena Text: Polish14351479Gemini 2.5 Pro
LMArena Text: multi-turn14221453Gemini 2.5 Pro
LMArena Text: longer queries14181449Gemini 2.5 Pro
LMArena Text: writing and language14071454Gemini 2.5 Pro
LMArena Text: entertainment and media14071441Gemini 2.5 Pro
LMArena Vision: EnglishNot measured1261Only Gemini 2.5 Pro measured
LMArena Vision: ChineseNot measured1300Only Gemini 2.5 Pro measured
LMArena Vision: reading textNot measured1270Only Gemini 2.5 Pro measured
LMArena Vision: humorNot measured1262Only Gemini 2.5 Pro measured
LMArena Vision: captioningNot measured1304Only Gemini 2.5 Pro measured
LMArena Vision: recognizing thingsNot measured1283Only Gemini 2.5 Pro measured
LMArena Vision: creative writingNot measured1265Only Gemini 2.5 Pro measured
LMArena DocumentNot measured1421Only Gemini 2.5 Pro measured
Neither model has a result on the other 10 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for DeepSeek V3.2 Exp and Gemini 2.5 Pro
BenchmarkDeepSeek V3.2 ExpGemini 2.5 ProLeader
Neither model has a result on 5 boards here

What each workload costs

Three fixed token mixes at each model's listed price, so per-token rates become a bill.

Chat turn

1K input + 500 output tokens

DeepSeek V3.2 Exp$0.0005Fits in one request
Gemini 2.5 Pro$0.0063Fits in one request

DeepSeek V3.2 Exp costs less

Repository review

50K input + 3K output tokens

DeepSeek V3.2 Exp$0.015Fits in one request
Gemini 2.5 Pro$0.092Fits in one request

DeepSeek V3.2 Exp costs less

Agent loop

200K input + 10K output tokens

DeepSeek V3.2 Exp$0.058Exceeds the context window
Gemini 2.5 Pro$0.350Fits in one request

DeepSeek V3.2 Exp costs less

Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.

Specification differences

The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.

Specifications of DeepSeek V3.2 Exp and Gemini 2.5 Pro
SpecificationDeepSeek V3.2 ExpGoogle Gemini 2.5 Pro
MakerDeepSeekGoogle
WeightsOpen weightsNo public weights listed
ReleasedSeptember 29, 2025June 17, 2025
Context window164K1.05M
Input $/Mtok$0.27$1.25
Output $/Mtok$0.41$10.00
Coding indexNot benchmarked19.1
Independent measurements behind it13
Cost per solved taskNot computable$0.64
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V3.2 Exp better than Gemini 2.5 Pro for coding?

Only Gemini 2.5 Pro has a coding score, 19.1 on the Vibe Coding Index; DeepSeek V3.2 Exp has none, so the two cannot be ranked on coding. Pick DeepSeek V3.2 Exp for output tokens 24.4x cheaper ($0.41 vs $10.00 per million) and work measured by LMArena WebDev or Terminal-Bench Hard (Artificial Analysis run), where it leads.

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

DeepSeek V3.2 Exp costs $0.27 per million input tokens and $0.41 per million output tokens; Gemini 2.5 Pro costs $1.25 and $10.00. Cost per solved task is known for Gemini 2.5 Pro only.

Which has the bigger context window, DeepSeek V3.2 Exp or Gemini 2.5 Pro?

Gemini 2.5 Pro, with 1.05M tokens against 164K.

How many benchmarks is the DeepSeek V3.2 Exp vs Gemini 2.5 Pro comparison based on?

52 benchmarks have a result for both models as of October 22, 2026. DeepSeek V3.2 Exp's coding index rests on 1 independent benchmark and Gemini 2.5 Pro's on 3. Where only one model has a result, the page says so instead of naming a winner.

Compare next