Enter opens the highlighted result. Esc closes.

Qwen3.8 27B vs MiMo-V2.5

Qwen3.8 27B leads MiMo-V2.5 on coding, 74.2 to 50.0 on the Vibe Coding Index, and MiMo-V2.5 costs 7.6x less per output token ($0.336 vs $2.55 per million). Updated October 22, 2026; 65 benchmarks have a result for both.

Replace Qwen3.8 27B Replace MiMo-V2.5
Qwen3.8 27BQwen
Coding
74.2
Overall
59.6
Output $/M
$2.55
55
MiMo-V2.5Xiaomi
Coding
50.0
Overall
50.7
Output $/M
$0.336
9
vs boards won of 65 shared, 1 tied

Qwen3.8 27B or MiMo-V2.5: which to pick

Pick Qwen3.8 27B for coding: it scores 74.2 on the Vibe Coding Index against 50.0 for MiMo-V2.5.

Pick MiMo-V2.5 for output tokens 7.6x cheaper ($0.336 vs $2.55 per million) and a lower cost per solved task ($0.01 against $1.26).

Model AQwen3.8 27BQwen74.2/1004 independent benchmarks$1.26 per solved taskModel BMiMo-V2.5Xiaomi50.0/1003 independent benchmarks$0.01 per solved task
Shared results
65
Qwen3.8 27B only
20
MiMo-V2.5 only
15
Like-for-like categories
7 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
Qwen3.8 27B 74.2 from 4 benchmarks, MiMo-V2.5 50.0 from 3 benchmarks.
Widest category
Agentic work, 19.3 points apart on average over 5 shared benchmarks.
Widest single gap
AA Agentic Index: Qwen3.8 27B 45.8 against MiMo-V2.5 15.8, 30.0 points apart on a 0 to 100 scale.
Output price
Qwen3.8 27B $2.55 against MiMo-V2.5 $0.336 per million output tokens, 7.6x apart.
Context window
Qwen3.8 27B takes 1M 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
CategoryQwen3.8 27BBarsMiMo-V2.5Ties
Coding102
Intelligence10
Agentic work50
Reasoning90
Math201 tie
Knowledge72
General215
All shared boards5591 tie
CodingLike-for-likeSix of these boards feed the Vibe Coding Index102
Coding benchmarks for Qwen3.8 27B and MiMo-V2.5
BenchmarkQwen3.8 27BMiMo-V2.5Leader
LiveBench coding86.8%Not measuredOnly Qwen3.8 27B measured
LMArena WebDev15931438Qwen3.8 27B
AA Coding Index68.156.8Qwen3.8 27B
LiveCodeBench (Vals AI)84.0%81.5%Qwen3.8 27B
SWE-bench Verified (Vals AI)86.0%71.0%Qwen3.8 27B
Vibe Code Bench (Vals AI)64.8%42.2%Qwen3.8 27B
Code migration (Vals AI)14.2%14.2%MiMo-V2.5
ProgramBench (Vals AI)0.0%Not measuredOnly Qwen3.8 27B measured
IOI (Vals AI)39.1%Not measuredOnly Qwen3.8 27B measured
Terminal-Bench 2.1 (Vals AI)58.4%60.7%MiMo-V2.5
SkillsBench (Vals AI)38.1%Not measuredOnly Qwen3.8 27B measured
CyberBench (Vals AI)67.8%Not measuredOnly Qwen3.8 27B measured
SciCode46.6%43.9%Qwen3.8 27B
Terminal-Bench Hard (Artificial Analysis run)Not measured41.7%Only MiMo-V2.5 measured
ALE-BenchNot measured514.0Only MiMo-V2.5 measured
LMArena Text: coding14831469Qwen3.8 27B
LMArena Text: software and IT14781463Qwen3.8 27B
LMArena WebDev: React16021425Qwen3.8 27B
LMArena WebDev: HTML15701429Qwen3.8 27B
LMArena WebDev: from an image1577Not measuredOnly Qwen3.8 27B measured
Neither model has a result on the other 19 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
Intelligence benchmarks for Qwen3.8 27B and MiMo-V2.5
BenchmarkQwen3.8 27BMiMo-V2.5Leader
AA Intelligence Index33.721.8Qwen3.8 27B
Epoch Capabilities Index151.7Not measuredOnly Qwen3.8 27B measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index50
Agentic work benchmarks for Qwen3.8 27B and MiMo-V2.5
BenchmarkQwen3.8 27BMiMo-V2.5Leader
AA Agentic Index45.815.8Qwen3.8 27B
LiveBench research59.4%Not measuredOnly Qwen3.8 27B measured
LiveBench browser32.0%Not measuredOnly Qwen3.8 27B measured
LiveBench long agentic27.9%Not measuredOnly Qwen3.8 27B measured
Finance Agent (Vals AI)48.5%36.7%Qwen3.8 27B
Tax Agent Bench (Vals AI)30.9%7.8%Qwen3.8 27B
Legal Research Bench (Vals AI)36.1%9.1%Qwen3.8 27B
Excel Modeling Benchmark (Vals AI)59.7%55.1%Qwen3.8 27B
Terminal-Bench-Science (Vals AI)0.0%Not measuredOnly Qwen3.8 27B measured
τ²-Bench TelecomNot measured90.6%Only MiMo-V2.5 measured
APEX-Agents47.5%Not measuredOnly Qwen3.8 27B measured
Neither model has a result on the other 30 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index90
Reasoning benchmarks for Qwen3.8 27B and MiMo-V2.5
BenchmarkQwen3.8 27BMiMo-V2.5Leader
GPQA Diamond88.9%81.6%Qwen3.8 27B
LiveBench reasoning80.0%Not measuredOnly Qwen3.8 27B measured
MMMU Pro (Vals AI)83.9%80.0%Qwen3.8 27B
GPQA Diamond (Artificial Analysis run)90.5%84.9%Qwen3.8 27B
Humanity's Last Exam (Artificial Analysis run)33.9%27.2%Qwen3.8 27B
ARC-AGI87.5%Not measuredOnly Qwen3.8 27B measured
ARC-AGI-242.4%Not measuredOnly Qwen3.8 27B measured
LMArena Text: hard prompts14591450Qwen3.8 27B
LMArena Text: expert14791459Qwen3.8 27B
LMArena Text: hard prompts in English14611461Qwen3.8 27B
LMArena Vision: diagrams12851269Qwen3.8 27B
LMArena Vision: homework12951262Qwen3.8 27B
Neither model has a result on the other 10 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index20
Math benchmarks for Qwen3.8 27B and MiMo-V2.5
BenchmarkQwen3.8 27BMiMo-V2.5Leader
ProofBench16.0%16.0%Tie
LMArena Text: math14551436Qwen3.8 27B
LMArena Text: mathematical work14581451Qwen3.8 27B
Neither model has a result on the other 10 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index72
Knowledge benchmarks for Qwen3.8 27B and MiMo-V2.5
BenchmarkQwen3.8 27BMiMo-V2.5Leader
MMLU Pro84.3%82.9%Qwen3.8 27B
LegalBench (Vals AI)82.4%78.9%Qwen3.8 27B
CorpFin (Vals AI)Not measured59.9%Only MiMo-V2.5 measured
TaxEval (Vals AI)70.8%71.8%MiMo-V2.5
MortgageTax (Vals AI)64.9%59.3%Qwen3.8 27B
MedCode (Vals AI)28.7%31.9%MiMo-V2.5
LMArena Text: science14671440Qwen3.8 27B
LMArena Text: medicine and healthcare14371432Qwen3.8 27B
LMArena Text: legal and government14471425Qwen3.8 27B
LMArena Text: business and finance14381432Qwen3.8 27B
Neither model has a result on the other 8 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index9 boards, none shared
Web design benchmarks for Qwen3.8 27B and MiMo-V2.5
BenchmarkQwen3.8 27BMiMo-V2.5Leader
Design Arena: websitesNot measured1265Only MiMo-V2.5 measured
Design Arena: UI componentsNot measured1255Only MiMo-V2.5 measured
Design Arena: game devNot measured1249Only MiMo-V2.5 measured
Design Arena: data vizNot measured1250Only MiMo-V2.5 measured
Design Arena: 3DNot measured1215Only MiMo-V2.5 measured
Design Arena: all code categoriesNot measured1260Only MiMo-V2.5 measured
Design Arena: SVGNot measured1179Only MiMo-V2.5 measured
Design Arena: ASCII artNot measured1140Only MiMo-V2.5 measured
LiveBench design75.9%Not measuredOnly Qwen3.8 27B measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index215
General benchmarks for Qwen3.8 27B and MiMo-V2.5
BenchmarkQwen3.8 27BMiMo-V2.5Leader
LiveBench data analysis76.6%Not measuredOnly Qwen3.8 27B measured
SAGE (Vals AI)52.4%43.3%Qwen3.8 27B
MedScribe (Vals AI)83.8%72.2%Qwen3.8 27B
Vals Multimodal IndexNot measured52.8%Only MiMo-V2.5 measured
IFBenchNot measured67.1%Only MiMo-V2.5 measured
Long context reasoning (AA-LCR)82.0%73.0%Qwen3.8 27B
DTBench88.0%Not measuredOnly Qwen3.8 27B measured
LMCA41.4%Not measuredOnly Qwen3.8 27B measured
Surface Evolver Bench45.0%Not measuredOnly Qwen3.8 27B measured
LMArena Text14401428Qwen3.8 27B
LMArena Text: instruction following14361435Qwen3.8 27B
LMArena Text: creative writing13831393MiMo-V2.5
LMArena Vision12691247Qwen3.8 27B
LMArena Text: English14481447Qwen3.8 27B
LMArena Text: non-English14281405Qwen3.8 27B
LMArena Text: Chinese15011469Qwen3.8 27B
LMArena Text: German14431421Qwen3.8 27B
LMArena Text: Japanese13871306Qwen3.8 27B
LMArena Text: Korean13841361Qwen3.8 27B
LMArena Text: French14651448Qwen3.8 27B
LMArena Text: Spanish14491414Qwen3.8 27B
LMArena Text: Russian14151396Qwen3.8 27B
LMArena Text: Polish14391413Qwen3.8 27B
LMArena Text: multi-turn14401445MiMo-V2.5
LMArena Text: longer queries14481445Qwen3.8 27B
LMArena Text: writing and language14051409MiMo-V2.5
LMArena Text: entertainment and media13931399MiMo-V2.5
LMArena Vision: English12671259Qwen3.8 27B
LMArena Vision: Chinese13261326MiMo-V2.5
LMArena Vision: reading text12811256Qwen3.8 27B
LMArena Vision: humor12351226Qwen3.8 27B
LMArena Vision: recognizing thingsNot measured1189Only MiMo-V2.5 measured
LMArena Vision: creative writing12371230Qwen3.8 27B
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 Qwen3.8 27B and MiMo-V2.5
BenchmarkQwen3.8 27BMiMo-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

Qwen3.8 27B$0.0017Fits in one request
MiMo-V2.5$0.0003Fits in one request

MiMo-V2.5 costs less

Repository review

50K input + 3K output tokens

Qwen3.8 27B$0.029Fits in one request
MiMo-V2.5$0.0094Fits in one request

MiMo-V2.5 costs less

Agent loop

200K input + 10K output tokens

Qwen3.8 27B$0.111Fits in one request
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 Qwen3.8 27B and MiMo-V2.5
SpecificationQwen3.8 27BXiaomi MiMo-V2.5
MakerQwenXiaomi
WeightsOpen weightsOpen weights
ReleasedAugust 14, 2026April 22, 2026
Context window1M1.05M
Input $/Mtok$0.425$0.168
Output $/Mtok$2.55$0.336
Coding index74.250.0
Independent measurements behind it43
Cost per solved task$1.26$0.01
SourceOpenRouter listingOpenRouter listing

Questions

Is Qwen3.8 27B better than MiMo-V2.5 for coding?

Pick Qwen3.8 27B for coding: it scores 74.2 on the Vibe Coding Index against 50.0 for MiMo-V2.5. Pick MiMo-V2.5 for output tokens 7.6x cheaper ($0.336 vs $2.55 per million) and a lower cost per solved task ($0.01 against $1.26).

Which is cheaper, Qwen3.8 27B or MiMo-V2.5?

Qwen3.8 27B costs $0.425 per million input tokens and $2.55 per million output tokens; MiMo-V2.5 costs $0.168 and $0.336. Per solved SWE-bench Verified task, Qwen3.8 27B costs $1.26 and MiMo-V2.5 $0.01.

Which has the bigger context window, Qwen3.8 27B or MiMo-V2.5?

MiMo-V2.5, with 1.05M tokens against 1M.

How many benchmarks is the Qwen3.8 27B vs MiMo-V2.5 comparison based on?

65 benchmarks have a result for both models as of October 22, 2026. Qwen3.8 27B's coding index rests on 4 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/qwen3-8-27b-vs-xiaomi-mimo-v2-5" title="Qwen3.8 27B 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/qwen3-8-27b-vs-xiaomi-mimo-v2-5">LLMBenchmark.ai, Qwen3.8 27B vs MiMo-V2.5</a></p>