LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

DeepSeek V3.2 vs GPT-4.1

Updated October 21, 2026. 61 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace DeepSeek V3.2 Replace GPT-4.1
DeepSeek V3.2DeepSeek
Coding
54.4
Overall
21.5
Output $/M
$0.42
56
GPT-4.1OpenAI
Coding
46.5
Overall
12.7
Output $/M
$8.00
5
vs boards won of 61 shared

Decision reading

Pick DeepSeek V3.2 for coding: it scores 54.4 on the Vibe Coding Index against 46.5 for GPT-4.1.

Pick GPT-4.1 when a solved task costs $0.37 against $0.99, or when the context window is larger, 1.05M against 164K tokens.

Model ADeepSeek V3.2DeepSeek54.4/1002 independent benchmarks$0.99 per solved taskModel BGPT-4.1OpenAI46.5/1002 independent benchmarks$0.37 per solved task
Shared results
61
DeepSeek V3.2 only
14
GPT-4.1 only
26
Like-for-like categories
8 of 9

Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.

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 54.4 from 2 benchmarks, GPT-4.1 46.5 from 2 benchmarks.
Widest category
Agentic work, 43.6 points apart on average over 1 shared benchmark.
Widest single gap
AIME 2025 (Artificial Analysis run): DeepSeek V3.2 92.0% against GPT-4.1 34.7%, 57.3 points apart on a 0 to 100 scale.
Output price
DeepSeek V3.2 $0.42 against GPT-4.1 $8.00 per million output tokens, 19.0x apart.
Context window
DeepSeek V3.2 takes 164K tokens, GPT-4.1 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 V3.2BarsGPT-4.1Ties
Coding60
Intelligence20
Agentic work10
Reasoning80
Math51
Knowledge74
Web design60
General210
All shared boards565
CodingLike-for-likeFour of these boards feed the Vibe Coding Index60
Coding benchmarks for DeepSeek V3.2 and GPT-4.1
BenchmarkDeepSeek V3.2GPT-4.1Leader
Aider polyglotNot measured52.4%Only GPT-4.1 measured
SWE-bench Verified70.0%39.6%DeepSeek V3.2
SWE-bench MultimodalNot measured31.1%Only GPT-4.1 measured
SWE-bench Multilingual59.0%Not measuredOnly DeepSeek V3.2 measured
LMArena WebDev1362Not measuredOnly DeepSeek V3.2 measured
AA Coding Index44.2Not measuredOnly DeepSeek V3.2 measured
LiveCodeBench (Vals AI)80.7%54.7%DeepSeek V3.2
SWE-bench Verified (Vals AI)67.6%Not measuredOnly DeepSeek V3.2 measured
Vibe Code Bench (Vals AI)5.1%Not measuredOnly DeepSeek V3.2 measured
LiveCodeBench (Artificial Analysis run)86.2%45.7%DeepSeek V3.2
Terminal-Bench Hard (Artificial Analysis run)35.6%13.6%DeepSeek V3.2
Terminal-Bench39.6%Not measuredOnly DeepSeek V3.2 measured
WeirdMLNot measured39.0%Only GPT-4.1 measured
ALE-BenchNot measured558.1Only GPT-4.1 measured
LMArena Text: coding14531390DeepSeek V3.2
LMArena Text: software and IT14491391DeepSeek V3.2
LMArena WebDev: React1350Not measuredOnly DeepSeek V3.2 measured
LMArena WebDev: HTML1371Not measuredOnly DeepSeek V3.2 measured
Neither model has a result on the other 20 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for DeepSeek V3.2 and GPT-4.1
BenchmarkDeepSeek V3.2GPT-4.1Leader
AA Intelligence Index21.512.7DeepSeek V3.2
Epoch Capabilities Index146.1136.7DeepSeek V3.2
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
Agentic work benchmarks for DeepSeek V3.2 and GPT-4.1
BenchmarkDeepSeek V3.2GPT-4.1Leader
τ²-Bench Telecom90.6%47.1%DeepSeek V3.2
APEX-Agents21.3%Not measuredOnly DeepSeek V3.2 measured
Vending-Bench 2$1,034Not measuredOnly DeepSeek V3.2 measured
Neither model has a result on the other 36 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index80
Reasoning benchmarks for DeepSeek V3.2 and GPT-4.1
BenchmarkDeepSeek V3.2GPT-4.1Leader
GPQA Diamond80.3%65.4%DeepSeek V3.2
MMMU Pro (Vals AI)Not measured72.4%Only GPT-4.1 measured
GPQA Diamond (Artificial Analysis run)84.0%66.6%DeepSeek V3.2
Humanity's Last Exam (Artificial Analysis run)24.6%4.2%DeepSeek V3.2
GPQA Diamond (Epoch AI run)Not measured66.9%Only GPT-4.1 measured
ARC-AGI57.0%5.5%DeepSeek V3.2
ARC-AGI-24.0%0.4%DeepSeek V3.2
SimpleBenchNot measured27.0%Only GPT-4.1 measured
Chess puzzlesNot measured3.0%Only GPT-4.1 measured
LMArena Text: hard prompts14341384DeepSeek V3.2
LMArena Text: expert14361364DeepSeek V3.2
LMArena Text: hard prompts in English14461393DeepSeek V3.2
LMArena Vision: diagramsNot measured1215Only GPT-4.1 measured
LMArena Vision: homeworkNot measured1248Only GPT-4.1 measured
Neither model has a result on the other 7 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index51
Math benchmarks for DeepSeek V3.2 and GPT-4.1
BenchmarkDeepSeek V3.2GPT-4.1Leader
AIME84.6%39.6%DeepSeek V3.2
MATH 500 (Vals AI)Not measured87.2%Only GPT-4.1 measured
MGSM (Vals AI)86.0%87.7%GPT-4.1
AA Math Index92.034.7DeepSeek V3.2
AIME 2025 (Artificial Analysis run)92.0%34.7%DeepSeek V3.2
OTIS Mock AIMENot measured38.3%Only GPT-4.1 measured
FrontierMath, tiers 1 to 3Not measured6.0%Only GPT-4.1 measured
MATH level 5Not measured83.0%Only GPT-4.1 measured
LMArena Text: math14351369DeepSeek V3.2
LMArena Text: mathematical work14401360DeepSeek V3.2
Neither model has a result on the other 3 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index74
Knowledge benchmarks for DeepSeek V3.2 and GPT-4.1
BenchmarkDeepSeek V3.2GPT-4.1Leader
MMLU Pro84.9%80.5%DeepSeek V3.2
LegalBench (Vals AI)76.1%83.1%GPT-4.1
CaseLaw (Vals AI)55.4%69.9%GPT-4.1
CorpFin (Vals AI)51.0%63.0%GPT-4.1
TaxEval (Vals AI)68.2%75.1%GPT-4.1
MortgageTax (Vals AI)Not measured65.9%Only GPT-4.1 measured
MedQA (Vals AI)93.9%91.2%DeepSeek V3.2
MMLU Pro (Artificial Analysis run)86.2%80.6%DeepSeek V3.2
SimpleQA VerifiedNot measured31.1%Only GPT-4.1 measured
GeoBenchNot measured72.0%Only GPT-4.1 measured
LMArena Text: science14461382DeepSeek V3.2
LMArena Text: medicine and healthcare14401379DeepSeek V3.2
LMArena Text: legal and government14251399DeepSeek V3.2
LMArena Text: business and finance14101369DeepSeek V3.2
Neither model has a result on the other 3 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index60
Web design benchmarks for DeepSeek V3.2 and GPT-4.1
BenchmarkDeepSeek V3.2GPT-4.1Leader
Design Arena: websites11751039DeepSeek V3.2
Design Arena: UI components11491000DeepSeek V3.2
Design Arena: game dev11321080DeepSeek V3.2
Design Arena: data viz11581104DeepSeek V3.2
Design Arena: 3D1133850DeepSeek V3.2
Design Arena: all code categories11641026DeepSeek V3.2
Design Arena: SVG1032Not measuredOnly DeepSeek V3.2 measured
Design Arena: ASCII art1082Not measuredOnly DeepSeek V3.2 measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index210
General benchmarks for DeepSeek V3.2 and GPT-4.1
BenchmarkDeepSeek V3.2GPT-4.1Leader
IFBench60.7%43.0%DeepSeek V3.2
Long context reasoning (AA-LCR)73.3%68.3%DeepSeek V3.2
DTBench85.6%68.3%DeepSeek V3.2
LMCA28.8%25.6%DeepSeek V3.2
Fiction.LiveBenchNot measured63.9%Only GPT-4.1 measured
CL-bench12.4%Not measuredOnly DeepSeek V3.2 measured
CL-bench Life7.4%Not measuredOnly DeepSeek V3.2 measured
LMArena Text14251383DeepSeek V3.2
LMArena Text: instruction following14131367DeepSeek V3.2
LMArena Text: creative writing14001363DeepSeek V3.2
LMArena VisionNot measured1210Only GPT-4.1 measured
LMArena Text: English14391392DeepSeek V3.2
LMArena Text: non-English14091370DeepSeek V3.2
LMArena Text: Chinese14551382DeepSeek V3.2
LMArena Text: German14131381DeepSeek V3.2
LMArena Text: Japanese13741319DeepSeek V3.2
LMArena Text: Korean13701338DeepSeek V3.2
LMArena Text: French14291382DeepSeek V3.2
LMArena Text: Spanish14241376DeepSeek V3.2
LMArena Text: Russian14241377DeepSeek V3.2
LMArena Text: Polish14191389DeepSeek V3.2
LMArena Text: multi-turn14271397DeepSeek V3.2
LMArena Text: longer queries14281385DeepSeek V3.2
LMArena Text: writing and language14071371DeepSeek V3.2
LMArena Text: entertainment and media13951365DeepSeek V3.2
LMArena Vision: EnglishNot measured1209Only GPT-4.1 measured
LMArena Vision: ChineseNot measured1226Only GPT-4.1 measured
LMArena Vision: reading textNot measured1218Only GPT-4.1 measured
LMArena Vision: humorNot measured1196Only GPT-4.1 measured
LMArena Vision: captioningNot measured1207Only GPT-4.1 measured
LMArena Vision: recognizing thingsNot measured1208Only GPT-4.1 measured
LMArena Vision: creative writingNot measured1210Only GPT-4.1 measured
Neither model has a result on the other 14 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for DeepSeek V3.2 and GPT-4.1
BenchmarkDeepSeek V3.2GPT-4.1Leader
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$0.0005Fits in one request
GPT-4.1$0.0060Fits in one request

DeepSeek V3.2 costs less

Repository review

50K input + 3K output tokens

DeepSeek V3.2$0.015Fits in one request
GPT-4.1$0.124Fits in one request

DeepSeek V3.2 costs less

Agent loop

200K input + 10K output tokens

DeepSeek V3.2$0.060Exceeds the context window
GPT-4.1$0.480Fits in one request

DeepSeek V3.2 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 and GPT-4.1
SpecificationDeepSeek V3.2OpenAI GPT-4.1
MakerDeepSeekOpenAI
WeightsOpen weightsNo public weights listed
ReleasedDecember 1, 2025April 14, 2025
Context window164K1.05M
Input $/Mtok$0.28$2.00
Output $/Mtok$0.42$8.00
Coding index54.446.5
Independent measurements behind it22
Cost per solved task$0.99$0.37
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V3.2 better than GPT-4.1 for coding?

Pick DeepSeek V3.2 for coding: it scores 54.4 on the Vibe Coding Index against 46.5 for GPT-4.1. Pick GPT-4.1 when a solved task costs $0.37 against $0.99, or when the context window is larger, 1.05M against 164K tokens.

Which is cheaper, DeepSeek V3.2 or GPT-4.1?

DeepSeek V3.2 costs $0.28 per million input tokens and $0.42 per million output tokens; GPT-4.1 costs $2.00 and $8.00. Per solved SWE-bench Verified task, DeepSeek V3.2 costs $0.99 and GPT-4.1 $0.37.

Which has the bigger context window, DeepSeek V3.2 or GPT-4.1?

GPT-4.1, with 1.05M tokens against 164K.

How many benchmarks is the DeepSeek V3.2 vs GPT-4.1 comparison based on?

61 benchmarks have a result for both models as of October 21, 2026. DeepSeek V3.2's coding index rests on 2 independent benchmarks and GPT-4.1's on 2. Where only one model has a result, the page says so instead of naming a winner.

Compare next

Embed this comparison on your site

Free under CC BY 4.0, while the source link stays visible. Preview the frame · Terms and other embeds

<iframe src="https://llmbenchmark.ai/embed/compare/deepseek-v3-2-vs-gpt-4-1" title="DeepSeek V3.2 vs GPT-4.1, LLMBenchmark.ai" width="100%" height="360" style="border:0;max-width:100%" loading="lazy"></iframe>
<p style="font:13px/1.4 sans-serif;margin:6px 0 0">Source: <a href="https://llmbenchmark.ai/compare/deepseek-v3-2-vs-gpt-4-1">LLMBenchmark.ai, DeepSeek V3.2 vs GPT-4.1</a></p>