Enter opens the highlighted result. Esc closes.

DeepSeek V3.2 Exp vs o3

o3 leads DeepSeek V3.2 Exp on the Artificial Analysis Intelligence Index, 20.2 to 16.6, and DeepSeek V3.2 Exp costs 19.5x less per output token ($0.41 vs $8.00 per million). Updated October 22, 2026; 53 benchmarks have a result for both.

Replace DeepSeek V3.2 Exp Replace o3
DeepSeek V3.2 ExpDeepSeek
Coding
No data
Overall
No data
Output $/M
$0.41
29
o3OpenAI
Coding
No data
Overall
51.1
Output $/M
$8.00
24
vs boards won of 53 shared

DeepSeek V3.2 Exp or o3: which to pick

Neither model has a coding score yet, so the two cannot be ranked on coding.

Pick DeepSeek V3.2 Exp for output tokens 19.5x cheaper ($0.41 vs $8.00 per million) and work measured by Aider polyglot or LiveCodeBench (Artificial Analysis run), where it leads.

Model ADeepSeek V3.2 ExpDeepSeekNot benchmarked1 independent benchmarkNo cost per solved taskModel Bo3OpenAINot benchmarkedNo coding benchmark$0.571 per solved task
Shared results
53
DeepSeek V3.2 Exp only
6
o3 only
49
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, o3 not benchmarked.
Widest category
Agentic work, 26.6 points apart on average over 2 shared benchmarks.
Widest single gap
τ²-Bench Telecom: DeepSeek V3.2 Exp 33.9% against o3 80.7%, 46.8 points apart on a 0 to 100 scale.
Output price
DeepSeek V3.2 Exp $0.41 against o3 $8.00 per million output tokens, 19.5x apart.
Context window
DeepSeek V3.2 Exp takes 164K tokens, o3 200K.

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 ExpBarso3Ties
Coding24
Intelligence02
Agentic work02
Reasoning33
Math13
Knowledge32
Web design41
General167
All shared boards2924
CodingLike-for-likeSix of these boards feed the Vibe Coding Index24
Coding benchmarks for DeepSeek V3.2 Exp and o3
BenchmarkDeepSeek V3.2 Expo3Leader
Aider polyglot74.2%81.3%o3
SWE-bench VerifiedNot measured58.4%Only o3 measured
SWE-bench MultimodalNot measured36.0%Only o3 measured
LMArena WebDev1272Not measuredOnly DeepSeek V3.2 Exp measured
LiveCodeBench (Vals AI)Not measured83.9%Only o3 measured
LiveCodeBench (Artificial Analysis run)78.9%80.8%o3
Terminal-Bench Hard (Artificial Analysis run)31.1%37.1%o3
GSO-BenchNot measured8.8%Only o3 measured
WeirdML39.5%52.4%o3
ALE-BenchNot measured933.5Only o3 measured
LMArena Text: coding14381407DeepSeek V3.2 Exp
LMArena Text: software and IT14381415DeepSeek V3.2 Exp
LMArena WebDev: HTML1281Not measuredOnly DeepSeek V3.2 Exp measured
SciCodeNot measured9.2%Only o3 measured
Neither model has a result on the other 25 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for DeepSeek V3.2 Exp and o3
BenchmarkDeepSeek V3.2 Expo3Leader
AA Intelligence Index16.620.2o3
Epoch Capabilities Index144.9146.7o3
Agentic workLike-for-likePublic boards, shown as published, kept out of the index02
Agentic work benchmarks for DeepSeek V3.2 Exp and o3
BenchmarkDeepSeek V3.2 Expo3Leader
τ²-Bench Telecom33.9%80.7%o3
METR time horizonsNot measured65.4%Only o3 measured
GDPvalNot measured30.8%Only o3 measured
DeepResearch BenchNot measured45.2%Only o3 measured
TheAgentCompany42.9%Not measuredOnly DeepSeek V3.2 Exp measured
OSWorldNot measured23.0%Only o3 measured
Berkeley Function Calling Leaderboard (BFCL)56.7%63.0%o3
BrowseComp-PlusNot measured65.9%Only o3 measured
Neither model has a result on the other 33 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index33
Reasoning benchmarks for DeepSeek V3.2 Exp and o3
BenchmarkDeepSeek V3.2 Expo3Leader
GPQA DiamondNot measured84.1%Only o3 measured
MMMU Pro (Vals AI)Not measured80.4%Only o3 measured
GPQA Diamond (Artificial Analysis run)79.7%82.7%o3
Humanity's Last Exam (Artificial Analysis run)14.9%20.1%o3
GPQA Diamond (Epoch AI run)Not measured81.8%Only o3 measured
ARC-AGINot measured60.8%Only o3 measured
ARC-AGI-2Not measured6.5%Only o3 measured
SimpleBenchNot measured53.1%Only o3 measured
Chess puzzlesNot measured37.0%Only o3 measured
Mystery game puzzlesNot measured29.0%Only o3 measured
VPCTNot measured52.0%Only o3 measured
LMArena Text: hard prompts14291402DeepSeek V3.2 Exp
LMArena Text: expert14221402DeepSeek V3.2 Exp
LMArena Text: hard prompts in English14441411DeepSeek V3.2 Exp
LMArena Vision: diagramsNot measured1220Only o3 measured
LMArena Vision: homeworkNot measured1234Only o3 measured
Kagi LLM Benchmark52.2%67.6%o3
Neither model has a result on the other 5 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index13
Math benchmarks for DeepSeek V3.2 Exp and o3
BenchmarkDeepSeek V3.2 Expo3Leader
AIMENot measured85.3%Only o3 measured
MATH 500 (Vals AI)Not measured94.6%Only o3 measured
MGSM (Vals AI)Not measured91.8%Only o3 measured
AA Math Index87.788.3o3
AIME 2025 (Artificial Analysis run)87.7%88.3%o3
OTIS Mock AIMENot measured84.4%Only o3 measured
FrontierMath, tiers 1 to 3Not measured33.3%Only o3 measured
MATH level 5Not measured97.8%Only o3 measured
LMArena Text: math14221426o3
LMArena Text: mathematical work14331423DeepSeek V3.2 Exp
Neither model has a result on the other 3 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index32
Knowledge benchmarks for DeepSeek V3.2 Exp and o3
BenchmarkDeepSeek V3.2 Expo3Leader
MMLU ProNot measured85.6%Only o3 measured
LegalBench (Vals AI)Not measured83.8%Only o3 measured
CorpFin (Vals AI)Not measured59.7%Only o3 measured
TaxEval (Vals AI)Not measured74.6%Only o3 measured
MortgageTax (Vals AI)Not measured65.7%Only o3 measured
MedQA (Vals AI)Not measured96.1%Only o3 measured
MedCode (Vals AI)Not measured47.3%Only o3 measured
MMLU Pro (Artificial Analysis run)85.0%85.3%o3
SimpleQA VerifiedNot measured49.4%Only o3 measured
GeoBenchNot measured74.0%Only o3 measured
LMArena Text: science14431429DeepSeek V3.2 Exp
LMArena Text: medicine and healthcare14681448DeepSeek V3.2 Exp
LMArena Text: legal and government14191428o3
LMArena Text: business and finance14121393DeepSeek V3.2 Exp
FACTS Benchmark SuiteNot measured44.4%Only o3 measured
Neither model has a result on the other 3 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index41
Web design benchmarks for DeepSeek V3.2 Exp and o3
BenchmarkDeepSeek V3.2 Expo3Leader
Design Arena: websites11781036DeepSeek V3.2 Exp
Design Arena: UI components11641017DeepSeek V3.2 Exp
Design Arena: game dev11441034DeepSeek V3.2 Exp
Design Arena: data viz11521200o3
Design Arena: 3D1148Not measuredOnly DeepSeek V3.2 Exp measured
Design Arena: all code categories11701020DeepSeek 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 index167
General benchmarks for DeepSeek V3.2 Exp and o3
BenchmarkDeepSeek V3.2 Expo3Leader
SAGE (Vals AI)Not measured41.8%Only o3 measured
MedScribe (Vals AI)Not measured76.7%Only o3 measured
IFBench54.1%71.4%o3
Long context reasoning (AA-LCR)72.3%74.7%o3
DTBench87.7%84.8%DeepSeek V3.2 Exp
LMCA29.1%39.7%o3
Fiction.LiveBench83.3%88.9%o3
Creative writing (Lech Mazur)Not measured83.9%Only o3 measured
CL-bench13.2%17.8%o3
CL-bench Life9.5%Not measuredOnly DeepSeek V3.2 Exp measured
LMArena Text14251410DeepSeek V3.2 Exp
LMArena Text: instruction following14011368DeepSeek V3.2 Exp
LMArena Text: creative writing14041359DeepSeek V3.2 Exp
LMArena VisionNot measured1214Only o3 measured
LMArena Text: English14411417DeepSeek V3.2 Exp
LMArena Text: non-English14041401DeepSeek V3.2 Exp
LMArena Text: Chinese14611437DeepSeek V3.2 Exp
LMArena Text: German14401420DeepSeek V3.2 Exp
LMArena Text: Japanese13311403o3
LMArena Text: Korean13721370DeepSeek V3.2 Exp
LMArena Text: French14341430DeepSeek V3.2 Exp
LMArena Text: Spanish14411395DeepSeek V3.2 Exp
LMArena Text: Russian14121406DeepSeek V3.2 Exp
LMArena Text: Polish14351438o3
LMArena Text: multi-turn14221405DeepSeek V3.2 Exp
LMArena Text: longer queries14181372DeepSeek V3.2 Exp
LMArena Text: writing and language14071372DeepSeek V3.2 Exp
LMArena Text: entertainment and media14071373DeepSeek V3.2 Exp
LMArena Vision: EnglishNot measured1221Only o3 measured
LMArena Vision: ChineseNot measured1237Only o3 measured
LMArena Vision: reading textNot measured1218Only o3 measured
LMArena Vision: humorNot measured1222Only o3 measured
LMArena Vision: captioningNot measured1210Only o3 measured
LMArena Vision: recognizing thingsNot measured1234Only o3 measured
LMArena Vision: creative writingNot measured1201Only o3 measured
Neither model has a result on the other 11 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 o3
BenchmarkDeepSeek V3.2 Expo3Leader
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
o3$0.0060Fits 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
o3$0.124Fits 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
o3$0.480Exceeds the context window

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 o3
SpecificationDeepSeek V3.2 ExpOpenAI o3
MakerDeepSeekOpenAI
WeightsOpen weightsNo public weights listed
ReleasedSeptember 29, 2025April 16, 2025
Context window164K200K
Input $/Mtok$0.27$2.00
Output $/Mtok$0.41$8.00
Coding indexNot benchmarkedNot benchmarked
Independent measurements behind it10
Cost per solved taskNot computable$0.571
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V3.2 Exp better than o3 for coding?

Neither model has a coding score yet, so the two cannot be ranked on coding. Pick DeepSeek V3.2 Exp for output tokens 19.5x cheaper ($0.41 vs $8.00 per million) and work measured by Aider polyglot or LiveCodeBench (Artificial Analysis run), where it leads.

Which is cheaper, DeepSeek V3.2 Exp or o3?

DeepSeek V3.2 Exp costs $0.27 per million input tokens and $0.41 per million output tokens; o3 costs $2.00 and $8.00. Cost per solved task is known for o3 only.

Which has the bigger context window, DeepSeek V3.2 Exp or o3?

o3, with 200K tokens against 164K.

How many benchmarks is the DeepSeek V3.2 Exp vs o3 comparison based on?

53 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 o3's on 0. Where only one model has a result, the page says so instead of naming a winner.

Compare next