Enter opens the highlighted result. Esc closes.

DeepSeek V3.2 vs MiMo-V2.5

MiMo-V2.5 leads DeepSeek V3.2 on coding, 50.0 to 30.0 on the Vibe Coding Index, and MiMo-V2.5 costs 2.6x less per output token ($0.336 vs $0.88 per million). Updated October 22, 2026; 55 benchmarks have a result for both.

Replace DeepSeek V3.2 Replace MiMo-V2.5
DeepSeek V3.2DeepSeek
Coding
30.0
Overall
37.9
Output $/M
$0.88
12
MiMo-V2.5Xiaomi
Coding
50.0
Overall
50.7
Output $/M
$0.336
42
vs boards won of 55 shared, 1 tied

DeepSeek V3.2 or MiMo-V2.5: which to pick

Pick MiMo-V2.5 for coding: it scores 50.0 on the Vibe Coding Index against 30.0 for DeepSeek V3.2.

Pick DeepSeek V3.2 for work measured by MMLU Pro or LMArena Text: science, where it leads.

Model ADeepSeek V3.2DeepSeek30.0/1003 independent benchmarks$0.99 per solved taskModel BMiMo-V2.5Xiaomi50.0/1003 independent benchmarks$0.01 per solved task
Shared results
55
DeepSeek V3.2 only
20
MiMo-V2.5 only
25
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 30.0 from 3 benchmarks, MiMo-V2.5 50.0 from 3 benchmarks.
Widest category
Web design, 12.4 points apart on average over 8 shared benchmarks.
Widest single gap
Vibe Code Bench (Vals AI): DeepSeek V3.2 5.1% against MiMo-V2.5 42.2%, 37.1 points apart on a 0 to 100 scale.
Output price
DeepSeek V3.2 $0.88 against MiMo-V2.5 $0.336 per million output tokens, 2.6x apart.
Context window
DeepSeek V3.2 takes 164K tokens, MiMo-V2.5 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 V3.2BarsMiMo-V2.5Ties
Coding010
Intelligence01
Agentic work001 tie
Reasoning06
Math02
Knowledge44
Web design08
General811
All shared boards12421 tie
CodingLike-for-likeSix of these boards feed the Vibe Coding Index010
Coding benchmarks for DeepSeek V3.2 and MiMo-V2.5
BenchmarkDeepSeek V3.2MiMo-V2.5Leader
SWE-bench Verified70.0%Not measuredOnly DeepSeek V3.2 measured
SWE-bench Multilingual59.0%Not measuredOnly DeepSeek V3.2 measured
LMArena WebDev13621438MiMo-V2.5
AA Coding Index44.256.8MiMo-V2.5
LiveCodeBench (Vals AI)80.7%81.5%MiMo-V2.5
SWE-bench Verified (Vals AI)67.6%71.0%MiMo-V2.5
Vibe Code Bench (Vals AI)5.1%42.2%MiMo-V2.5
Code migration (Vals AI)Not measured14.2%Only MiMo-V2.5 measured
Terminal-Bench 2.1 (Vals AI)Not measured60.7%Only MiMo-V2.5 measured
LiveCodeBench (Artificial Analysis run)86.2%Not measuredOnly DeepSeek V3.2 measured
SciCodeNot measured43.9%Only MiMo-V2.5 measured
Terminal-Bench Hard (Artificial Analysis run)35.6%41.7%MiMo-V2.5
Terminal-Bench39.6%Not measuredOnly DeepSeek V3.2 measured
ALE-BenchNot measured514.0Only MiMo-V2.5 measured
LMArena Text: coding14531469MiMo-V2.5
LMArena Text: software and IT14491463MiMo-V2.5
LMArena WebDev: React13501425MiMo-V2.5
LMArena WebDev: HTML13711429MiMo-V2.5
Neither model has a result on the other 21 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
Intelligence benchmarks for DeepSeek V3.2 and MiMo-V2.5
BenchmarkDeepSeek V3.2MiMo-V2.5Leader
AA Intelligence Index21.521.8MiMo-V2.5
Epoch Capabilities Index146.1Not measuredOnly DeepSeek V3.2 measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index00
Agentic work benchmarks for DeepSeek V3.2 and MiMo-V2.5
BenchmarkDeepSeek V3.2MiMo-V2.5Leader
AA Agentic IndexNot measured15.8Only MiMo-V2.5 measured
Finance Agent (Vals AI)Not measured36.7%Only MiMo-V2.5 measured
Tax Agent Bench (Vals AI)Not measured7.8%Only MiMo-V2.5 measured
Legal Research Bench (Vals AI)Not measured9.1%Only MiMo-V2.5 measured
Excel Modeling Benchmark (Vals AI)Not measured55.1%Only MiMo-V2.5 measured
τ²-Bench Telecom90.6%90.6%Tie
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 33 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index06
Reasoning benchmarks for DeepSeek V3.2 and MiMo-V2.5
BenchmarkDeepSeek V3.2MiMo-V2.5Leader
GPQA Diamond80.3%81.6%MiMo-V2.5
MMMU Pro (Vals AI)Not measured80.0%Only MiMo-V2.5 measured
GPQA Diamond (Artificial Analysis run)84.0%84.9%MiMo-V2.5
Humanity's Last Exam (Artificial Analysis run)24.6%27.2%MiMo-V2.5
ARC-AGI57.0%Not measuredOnly DeepSeek V3.2 measured
ARC-AGI-24.0%Not measuredOnly DeepSeek V3.2 measured
LMArena Text: hard prompts14341450MiMo-V2.5
LMArena Text: expert14361459MiMo-V2.5
LMArena Text: hard prompts in English14461461MiMo-V2.5
LMArena Vision: diagramsNot measured1269Only MiMo-V2.5 measured
LMArena Vision: homeworkNot measured1262Only MiMo-V2.5 measured
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 and MiMo-V2.5
BenchmarkDeepSeek V3.2MiMo-V2.5Leader
AIME84.6%Not measuredOnly DeepSeek V3.2 measured
MGSM (Vals AI)86.0%Not measuredOnly DeepSeek V3.2 measured
AA Math Index92.0Not measuredOnly DeepSeek V3.2 measured
AIME 2025 (Artificial Analysis run)92.0%Not measuredOnly DeepSeek V3.2 measured
ProofBenchNot measured16.0%Only MiMo-V2.5 measured
LMArena Text: math14351436MiMo-V2.5
LMArena Text: mathematical work14401451MiMo-V2.5
Neither model has a result on the other 6 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index44
Knowledge benchmarks for DeepSeek V3.2 and MiMo-V2.5
BenchmarkDeepSeek V3.2MiMo-V2.5Leader
MMLU Pro84.9%82.9%DeepSeek V3.2
LegalBench (Vals AI)76.1%78.9%MiMo-V2.5
CaseLaw (Vals AI)55.4%Not measuredOnly DeepSeek V3.2 measured
CorpFin (Vals AI)51.0%59.9%MiMo-V2.5
TaxEval (Vals AI)68.2%71.8%MiMo-V2.5
MortgageTax (Vals AI)Not measured59.3%Only MiMo-V2.5 measured
MedQA (Vals AI)93.9%Not measuredOnly DeepSeek V3.2 measured
MedCode (Vals AI)Not measured31.9%Only MiMo-V2.5 measured
MMLU Pro (Artificial Analysis run)86.2%Not measuredOnly DeepSeek V3.2 measured
LMArena Text: science14461440DeepSeek V3.2
LMArena Text: medicine and healthcare14401432DeepSeek V3.2
LMArena Text: legal and government14251425DeepSeek V3.2
LMArena Text: business and finance14101432MiMo-V2.5
Neither model has a result on the other 5 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index08
Web design benchmarks for DeepSeek V3.2 and MiMo-V2.5
BenchmarkDeepSeek V3.2MiMo-V2.5Leader
Design Arena: websites11751265MiMo-V2.5
Design Arena: UI components11461255MiMo-V2.5
Design Arena: game dev11311249MiMo-V2.5
Design Arena: data viz11581250MiMo-V2.5
Design Arena: 3D11321215MiMo-V2.5
Design Arena: all code categories11631260MiMo-V2.5
Design Arena: SVG10321179MiMo-V2.5
Design Arena: ASCII art10831140MiMo-V2.5
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index811
General benchmarks for DeepSeek V3.2 and MiMo-V2.5
BenchmarkDeepSeek V3.2MiMo-V2.5Leader
SAGE (Vals AI)Not measured43.3%Only MiMo-V2.5 measured
MedScribe (Vals AI)Not measured72.2%Only MiMo-V2.5 measured
Vals Multimodal IndexNot measured52.8%Only MiMo-V2.5 measured
IFBench60.7%67.1%MiMo-V2.5
Long context reasoning (AA-LCR)73.3%73.0%DeepSeek V3.2
DTBench85.6%Not measuredOnly DeepSeek V3.2 measured
LMCA28.8%Not measuredOnly DeepSeek V3.2 measured
CL-bench12.4%Not measuredOnly DeepSeek V3.2 measured
CL-bench Life7.4%Not measuredOnly DeepSeek V3.2 measured
LMArena Text14251428MiMo-V2.5
LMArena Text: instruction following14131435MiMo-V2.5
LMArena Text: creative writing14001393DeepSeek V3.2
LMArena VisionNot measured1247Only MiMo-V2.5 measured
LMArena Text: English14391447MiMo-V2.5
LMArena Text: non-English14091405DeepSeek V3.2
LMArena Text: Chinese14551469MiMo-V2.5
LMArena Text: German14131421MiMo-V2.5
LMArena Text: Japanese13741306DeepSeek V3.2
LMArena Text: Korean13701361DeepSeek V3.2
LMArena Text: French14291448MiMo-V2.5
LMArena Text: Spanish14241414DeepSeek V3.2
LMArena Text: Russian14241396DeepSeek V3.2
LMArena Text: Polish14191413DeepSeek V3.2
LMArena Text: multi-turn14271445MiMo-V2.5
LMArena Text: longer queries14281445MiMo-V2.5
LMArena Text: writing and language14071409MiMo-V2.5
LMArena Text: entertainment and media13951399MiMo-V2.5
LMArena Vision: EnglishNot measured1259Only MiMo-V2.5 measured
LMArena Vision: ChineseNot measured1326Only MiMo-V2.5 measured
LMArena Vision: reading textNot measured1256Only MiMo-V2.5 measured
LMArena Vision: humorNot measured1226Only MiMo-V2.5 measured
LMArena Vision: recognizing thingsNot measured1189Only MiMo-V2.5 measured
LMArena Vision: creative writingNot measured1230Only MiMo-V2.5 measured
Neither model has a result on the other 13 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 MiMo-V2.5
BenchmarkDeepSeek V3.2MiMo-V2.5Leader
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$0.0007Fits in one request
MiMo-V2.5$0.0003Fits in one request

MiMo-V2.5 costs less

Repository review

50K input + 3K output tokens

DeepSeek V3.2$0.017Fits in one request
MiMo-V2.5$0.0094Fits in one request

MiMo-V2.5 costs less

Agent loop

200K input + 10K output tokens

DeepSeek V3.2$0.067Exceeds the context window
MiMo-V2.5$0.037Fits in one request

MiMo-V2.5 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 MiMo-V2.5
SpecificationDeepSeek V3.2Xiaomi MiMo-V2.5
MakerDeepSeekXiaomi
WeightsOpen weightsOpen weights
ReleasedDecember 1, 2025April 22, 2026
Context window164K1.05M
Input $/Mtok$0.29$0.168
Output $/Mtok$0.88$0.336
Coding index30.050.0
Independent measurements behind it33
Cost per solved task$0.99$0.01
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V3.2 better than MiMo-V2.5 for coding?

Pick MiMo-V2.5 for coding: it scores 50.0 on the Vibe Coding Index against 30.0 for DeepSeek V3.2. Pick DeepSeek V3.2 for work measured by MMLU Pro or LMArena Text: science, where it leads.

Which is cheaper, DeepSeek V3.2 or MiMo-V2.5?

DeepSeek V3.2 costs $0.29 per million input tokens and $0.88 per million output tokens; MiMo-V2.5 costs $0.168 and $0.336. Per solved SWE-bench Verified task, DeepSeek V3.2 costs $0.99 and MiMo-V2.5 $0.01.

Which has the bigger context window, DeepSeek V3.2 or MiMo-V2.5?

MiMo-V2.5, with 1.05M tokens against 164K.

How many benchmarks is the DeepSeek V3.2 vs MiMo-V2.5 comparison based on?

55 benchmarks have a result for both models as of October 22, 2026. DeepSeek V3.2's coding index rests on 3 independent benchmarks and MiMo-V2.5's on 3. 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-xiaomi-mimo-v2-5" title="DeepSeek V3.2 vs MiMo-V2.5, 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-xiaomi-mimo-v2-5">LLMBenchmark.ai, DeepSeek V3.2 vs MiMo-V2.5</a></p>