LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

R1 0528 vs DeepSeek V3.2 Exp

Updated October 21, 2026. 49 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace R1 0528 Replace DeepSeek V3.2 Exp
R1 0528DeepSeek
Coding
71.4
Overall
13.1
Output $/M
$2.15
11
DeepSeek V3.2 ExpDeepSeek
Coding
55.0
Overall
16.6
Output $/M
$0.41
38
vs boards won of 49 shared

Decision reading

Pick R1 0528 for coding: it scores 71.4 on the Vibe Coding Index against 55.0 for DeepSeek V3.2 Exp, though on 1 benchmark against 2.

Pick DeepSeek V3.2 Exp when output tokens cost 5.2x less, or when what matters is Aider polyglot or LiveCodeBench (Artificial Analysis run), where it leads.

Model AR1 0528DeepSeek71.4/1001 independent benchmarkNo cost per solved taskModel BDeepSeek V3.2 ExpDeepSeek55.0/1002 independent benchmarksNo cost per solved task
Shared results
49
R1 0528 only
10
DeepSeek V3.2 Exp only
8
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
R1 0528 71.4 from 1 benchmark, DeepSeek V3.2 Exp 55.0 from 2 benchmarks.
Widest category
Math, 7.5 points apart on average over 4 shared benchmarks.
Widest single gap
Long context reasoning (AA-LCR): R1 0528 55.7% against DeepSeek V3.2 Exp 72.3%, 16.7 points apart on a 0 to 100 scale.
Output price
R1 0528 $2.15 against DeepSeek V3.2 Exp $0.41 per million output tokens, 5.2x apart.
Context window
Both take 164K tokens.

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
CategoryR1 0528BarsDeepSeek V3.2 ExpTies
Coding15
Intelligence01
Agentic work10
Reasoning23
Math04
Knowledge05
Web design25
General515
All shared boards1138
CodingLike-for-likeFour of these boards feed the Vibe Coding Index15
Coding benchmarks for R1 0528 and DeepSeek V3.2 Exp
BenchmarkR1 0528DeepSeek V3.2 ExpLeader
Aider polyglot71.4%74.2%DeepSeek V3.2 Exp
LMArena WebDevNot measured1272Only DeepSeek V3.2 Exp measured
LiveCodeBench (Artificial Analysis run)77.0%78.9%DeepSeek V3.2 Exp
Terminal-Bench Hard (Artificial Analysis run)15.9%31.1%DeepSeek V3.2 Exp
WeirdML41.6%39.5%R1 0528
ALE-Bench804.1Not measuredOnly R1 0528 measured
LMArena Text: coding14271438DeepSeek V3.2 Exp
LMArena Text: software and IT14351438DeepSeek V3.2 Exp
LMArena WebDev: HTMLNot measured1281Only DeepSeek V3.2 Exp measured
Neither model has a result on the other 29 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
Intelligence benchmarks for R1 0528 and DeepSeek V3.2 Exp
BenchmarkR1 0528DeepSeek V3.2 ExpLeader
AA Intelligence Index13.116.6DeepSeek V3.2 Exp
Epoch Capabilities IndexNot measured144.9Only DeepSeek V3.2 Exp measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
Agentic work benchmarks for R1 0528 and DeepSeek V3.2 Exp
BenchmarkR1 0528DeepSeek V3.2 ExpLeader
τ²-Bench Telecom36.5%33.9%R1 0528
METR time horizons53.8%Not measuredOnly R1 0528 measured
DeepResearch Bench35.1%Not measuredOnly R1 0528 measured
TheAgentCompanyNot measured42.9%Only DeepSeek V3.2 Exp measured
Neither model has a result on the other 35 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index23
Reasoning benchmarks for R1 0528 and DeepSeek V3.2 Exp
BenchmarkR1 0528DeepSeek V3.2 ExpLeader
GPQA Diamond (Artificial Analysis run)81.3%79.7%R1 0528
Humanity's Last Exam (Artificial Analysis run)15.8%14.9%R1 0528
GPQA Diamond (Epoch AI run)76.3%Not measuredOnly R1 0528 measured
ARC-AGI21.2%Not measuredOnly R1 0528 measured
ARC-AGI-21.1%Not measuredOnly R1 0528 measured
SimpleBench40.8%Not measuredOnly R1 0528 measured
LMArena Text: hard prompts14151429DeepSeek V3.2 Exp
LMArena Text: expert13941422DeepSeek V3.2 Exp
LMArena Text: hard prompts in English14221444DeepSeek V3.2 Exp
Neither model has a result on the other 12 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index04
Math benchmarks for R1 0528 and DeepSeek V3.2 Exp
BenchmarkR1 0528DeepSeek V3.2 ExpLeader
AA Math Index76.087.7DeepSeek V3.2 Exp
AIME 2025 (Artificial Analysis run)76.0%87.7%DeepSeek V3.2 Exp
OTIS Mock AIME66.4%Not measuredOnly R1 0528 measured
MATH level 596.6%Not measuredOnly R1 0528 measured
LMArena Text: math14001422DeepSeek V3.2 Exp
LMArena Text: mathematical work14151433DeepSeek V3.2 Exp
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index05
Knowledge benchmarks for R1 0528 and DeepSeek V3.2 Exp
BenchmarkR1 0528DeepSeek V3.2 ExpLeader
MMLU Pro (Artificial Analysis run)84.9%85.0%DeepSeek V3.2 Exp
LMArena Text: science14301443DeepSeek V3.2 Exp
LMArena Text: medicine and healthcare14451468DeepSeek V3.2 Exp
LMArena Text: legal and government14191419DeepSeek V3.2 Exp
LMArena Text: business and finance14021412DeepSeek V3.2 Exp
Neither model has a result on the other 12 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index25
Web design benchmarks for R1 0528 and DeepSeek V3.2 Exp
BenchmarkR1 0528DeepSeek V3.2 ExpLeader
Design Arena: websites11491179DeepSeek V3.2 Exp
Design Arena: UI components11021166DeepSeek V3.2 Exp
Design Arena: game dev10981145DeepSeek V3.2 Exp
Design Arena: data viz11821152R1 0528
Design Arena: 3D11111148DeepSeek V3.2 Exp
Design Arena: all code categories11391171DeepSeek V3.2 Exp
Design Arena: SVG10381031R1 0528
Neither model has a result on the other 2 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index515
General benchmarks for R1 0528 and DeepSeek V3.2 Exp
BenchmarkR1 0528DeepSeek V3.2 ExpLeader
IFBench39.6%54.1%DeepSeek V3.2 Exp
Long context reasoning (AA-LCR)55.7%72.3%DeepSeek V3.2 Exp
DTBenchNot measured87.7%Only DeepSeek V3.2 Exp measured
LMCANot measured29.1%Only DeepSeek V3.2 Exp measured
Fiction.LiveBench75.0%83.3%DeepSeek V3.2 Exp
Creative writing (Lech Mazur)81.9%Not measuredOnly R1 0528 measured
CL-benchNot measured13.2%Only DeepSeek V3.2 Exp measured
CL-bench LifeNot measured9.5%Only DeepSeek V3.2 Exp measured
LMArena Text14281425R1 0528
LMArena Text: instruction following13821401DeepSeek V3.2 Exp
LMArena Text: creative writing14051404R1 0528
LMArena Text: English14371441DeepSeek V3.2 Exp
LMArena Text: non-English14121404R1 0528
LMArena Text: Chinese14421461DeepSeek V3.2 Exp
LMArena Text: German14041440DeepSeek V3.2 Exp
LMArena Text: Japanese13911331R1 0528
LMArena Text: Korean13601372DeepSeek V3.2 Exp
LMArena Text: French14171434DeepSeek V3.2 Exp
LMArena Text: Spanish14111441DeepSeek V3.2 Exp
LMArena Text: Russian14231412R1 0528
LMArena Text: Polish14281435DeepSeek V3.2 Exp
LMArena Text: multi-turn14051422DeepSeek V3.2 Exp
LMArena Text: longer queries13911418DeepSeek V3.2 Exp
LMArena Text: writing and language14021407DeepSeek V3.2 Exp
LMArena Text: entertainment and media14001407DeepSeek V3.2 Exp
Neither model has a result on the other 21 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for R1 0528 and DeepSeek V3.2 Exp
BenchmarkR1 0528DeepSeek V3.2 ExpLeader
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

R1 0528$0.0016Fits in one request
DeepSeek V3.2 Exp$0.0005Fits in one request

DeepSeek V3.2 Exp costs less

Repository review

50K input + 3K output tokens

R1 0528$0.031Fits in one request
DeepSeek V3.2 Exp$0.015Fits in one request

DeepSeek V3.2 Exp costs less

Agent loop

200K input + 10K output tokens

R1 0528$0.121Exceeds the context window
DeepSeek V3.2 Exp$0.058Exceeds 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 R1 0528 and DeepSeek V3.2 Exp
SpecificationDeepSeek R1 0528DeepSeek V3.2 Exp
MakerDeepSeekDeepSeek
WeightsOpen weightsOpen weights
ReleasedMay 28, 2025September 29, 2025
Context window164K164K
Input $/Mtok$0.50$0.27
Output $/Mtok$2.15$0.41
Coding index71.455.0
Independent measurements behind it12
Cost per solved taskNot computableNot computable
SourceOpenRouter listingOpenRouter listing

Questions

Is R1 0528 better than DeepSeek V3.2 Exp for coding?

Pick R1 0528 for coding: it scores 71.4 on the Vibe Coding Index against 55.0 for DeepSeek V3.2 Exp, though on 1 benchmark against 2. Pick DeepSeek V3.2 Exp when output tokens cost 5.2x less, or when what matters is Aider polyglot or LiveCodeBench (Artificial Analysis run), where it leads.

Which is cheaper, R1 0528 or DeepSeek V3.2 Exp?

R1 0528 costs $0.50 per million input tokens and $2.15 per million output tokens; DeepSeek V3.2 Exp costs $0.27 and $0.41. Cost per solved task is known for neither of them.

Which has the bigger context window, R1 0528 or DeepSeek V3.2 Exp?

Neither: both take 164K tokens.

How many benchmarks is the R1 0528 vs DeepSeek V3.2 Exp comparison based on?

49 benchmarks have a result for both models as of October 21, 2026. R1 0528's coding index rests on 1 independent benchmark and DeepSeek V3.2 Exp's on 2. Where only one model has a result, the page says so instead of naming a winner.

Compare next