Enter opens the highlighted result. Esc closes.

Qwen3.5-Flash vs Qwen3.8 27B

Qwen3.8 27B leads Qwen3.5-Flash on coding, 74.1 to 41.1 on the Vibe Coding Index, and Qwen3.5-Flash costs 9.8x less per output token ($0.26 vs $2.55 per million). Updated October 23, 2026; 45 benchmarks have a result for both.

Replace Qwen3.5-Flash Replace Qwen3.8 27B
Qwen3.5-FlashQwen
Coding
41.1
Overall
50.1
Output $/M
$0.26
4
Qwen3.8 27BQwen
Coding
74.1
Overall
59.5
Output $/M
$2.55
41
vs boards won of 45 shared

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

Pick Qwen3.8 27B for coding: it scores 74.1 on the Vibe Coding Index against 41.1 for Qwen3.5-Flash.

Pick Qwen3.5-Flash for output tokens 9.8x cheaper ($0.26 vs $2.55 per million) and a lower cost per solved task ($0.40 against $1.26).

Model AQwen3.5-FlashQwen41.1/1002 independent benchmarks$0.40 per solved taskModel BQwen3.8 27BQwen74.1/1004 independent benchmarks$1.26 per solved task
Shared results
45
Qwen3.5-Flash only
10
Qwen3.8 27B only
40
Like-for-like categories
6 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
Qwen3.5-Flash 41.1 from 2 benchmarks, Qwen3.8 27B 74.1 from 4 benchmarks.
Widest category
Coding, 28.9 points apart on average over 7 shared benchmarks.
Widest single gap
LMArena WebDev: React: Qwen3.5-Flash 1229 against Qwen3.8 27B 1600, 61.8 points apart on a 0 to 100 scale.
Output price
Qwen3.5-Flash $0.26 against Qwen3.8 27B $2.55 per million output tokens, 9.8x apart.
Context window
Both take 1M 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
CategoryQwen3.5-FlashBarsQwen3.8 27BTies
Coding07
Intelligence01
Reasoning05
Math02
Knowledge45
General021
All shared boards441
CodingLike-for-likeSix of these boards feed the Vibe Coding Index07
Coding benchmarks for Qwen3.5-Flash and Qwen3.8 27B
BenchmarkQwen3.5-FlashQwen3.8 27BLeader
LiveBench codingNot measured86.8%Only Qwen3.8 27B measured
LMArena WebDev12441591Qwen3.8 27B
AA Coding IndexNot measured68.1Only Qwen3.8 27B measured
LiveCodeBench (Vals AI)83.3%84.0%Qwen3.8 27B
SWE-bench Verified (Vals AI)64.4%86.0%Qwen3.8 27B
Vibe Code Bench (Vals AI)Not measured64.8%Only Qwen3.8 27B measured
Code migration (Vals AI)Not measured14.2%Only Qwen3.8 27B measured
ProgramBench (Vals AI)Not measured0.0%Only Qwen3.8 27B measured
IOI (Vals AI)Not measured39.1%Only Qwen3.8 27B measured
Terminal-Bench 2.1 (Vals AI)Not measured58.4%Only Qwen3.8 27B measured
SkillsBench (Vals AI)Not measured38.1%Only Qwen3.8 27B measured
CyberBench (Vals AI)Not measured67.8%Only Qwen3.8 27B measured
SciCodeNot measured46.6%Only Qwen3.8 27B measured
ALE-Bench265.9Not measuredOnly Qwen3.5-Flash measured
LMArena Text: coding14121483Qwen3.8 27B
LMArena Text: software and IT14161477Qwen3.8 27B
LMArena WebDev: React12291600Qwen3.8 27B
LMArena WebDev: HTML12911569Qwen3.8 27B
LMArena WebDev: from an imageNot measured1577Only Qwen3.8 27B measured
Neither model has a result on the other 24 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
Intelligence benchmarks for Qwen3.5-Flash and Qwen3.8 27B
BenchmarkQwen3.5-FlashQwen3.8 27BLeader
AA Intelligence IndexNot measured33.7Only Qwen3.8 27B measured
Epoch Capabilities Index143.5151.7Qwen3.8 27B
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index10 boards, none shared
Agentic work benchmarks for Qwen3.5-Flash and Qwen3.8 27B
BenchmarkQwen3.5-FlashQwen3.8 27BLeader
AA Agentic IndexNot measured45.8Only Qwen3.8 27B measured
LiveBench researchNot measured59.4%Only Qwen3.8 27B measured
LiveBench browserNot measured32.0%Only Qwen3.8 27B measured
LiveBench long agenticNot measured27.9%Only Qwen3.8 27B measured
Finance Agent (Vals AI)Not measured48.5%Only Qwen3.8 27B measured
Tax Agent Bench (Vals AI)Not measured30.9%Only Qwen3.8 27B measured
Legal Research Bench (Vals AI)Not measured36.1%Only Qwen3.8 27B measured
Excel Modeling Benchmark (Vals AI)Not measured59.7%Only Qwen3.8 27B measured
Terminal-Bench-Science (Vals AI)Not measured0.0%Only Qwen3.8 27B measured
APEX-AgentsNot measured47.5%Only Qwen3.8 27B measured
Neither model has a result on the other 32 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index05
Reasoning benchmarks for Qwen3.5-Flash and Qwen3.8 27B
BenchmarkQwen3.5-FlashQwen3.8 27BLeader
GPQA Diamond82.8%88.9%Qwen3.8 27B
LiveBench reasoningNot measured80.0%Only Qwen3.8 27B measured
MMMU Pro (Vals AI)81.9%83.9%Qwen3.8 27B
GPQA Diamond (Artificial Analysis run)Not measured90.5%Only Qwen3.8 27B measured
Humanity's Last Exam (Artificial Analysis run)Not measured33.9%Only Qwen3.8 27B measured
GPQA Diamond (Epoch AI run)82.3%Not measuredOnly Qwen3.5-Flash measured
ARC-AGINot measured87.5%Only Qwen3.8 27B measured
ARC-AGI-2Not measured42.4%Only Qwen3.8 27B measured
Chess puzzles14.0%Not measuredOnly Qwen3.5-Flash measured
Mystery game puzzles20.0%Not measuredOnly Qwen3.5-Flash measured
LMArena Text: hard prompts14031459Qwen3.8 27B
LMArena Text: expert14071480Qwen3.8 27B
LMArena Text: hard prompts in English14091462Qwen3.8 27B
LMArena Vision: diagramsNot measured1285Only Qwen3.8 27B measured
LMArena Vision: homeworkNot measured1295Only Qwen3.8 27B measured
Neither model has a result on the other 8 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for Qwen3.5-Flash and Qwen3.8 27B
BenchmarkQwen3.5-FlashQwen3.8 27BLeader
AIME92.5%Not measuredOnly Qwen3.5-Flash measured
OTIS Mock AIME84.4%Not measuredOnly Qwen3.5-Flash measured
FrontierMath, tiers 1 to 318.2%Not measuredOnly Qwen3.5-Flash measured
ProofBenchNot measured16.0%Only Qwen3.8 27B measured
LMArena Text: math14071455Qwen3.8 27B
LMArena Text: mathematical work14131458Qwen3.8 27B
Neither model has a result on the other 8 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index45
Knowledge benchmarks for Qwen3.5-Flash and Qwen3.8 27B
BenchmarkQwen3.5-FlashQwen3.8 27BLeader
MMLU Pro84.1%84.3%Qwen3.8 27B
LegalBench (Vals AI)84.3%82.4%Qwen3.5-Flash
CaseLaw (Vals AI)56.0%Not measuredOnly Qwen3.5-Flash measured
CorpFin (Vals AI)63.6%Not measuredOnly Qwen3.5-Flash measured
TaxEval (Vals AI)72.2%70.8%Qwen3.5-Flash
MortgageTax (Vals AI)67.4%64.9%Qwen3.5-Flash
MedCode (Vals AI)33.0%28.7%Qwen3.5-Flash
SimpleQA Verified20.3%Not measuredOnly Qwen3.5-Flash measured
LMArena Text: science14151468Qwen3.8 27B
LMArena Text: medicine and healthcare14071438Qwen3.8 27B
LMArena Text: legal and government13971447Qwen3.8 27B
LMArena Text: business and finance13941438Qwen3.8 27B
Neither model has a result on the other 8 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index1 board, none shared
Web design benchmarks for Qwen3.5-Flash and Qwen3.8 27B
BenchmarkQwen3.5-FlashQwen3.8 27BLeader
LiveBench designNot measured75.9%Only Qwen3.8 27B measured
Neither model has a result on the other 8 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index021
General benchmarks for Qwen3.5-Flash and Qwen3.8 27B
BenchmarkQwen3.5-FlashQwen3.8 27BLeader
LiveBench data analysisNot measured76.6%Only Qwen3.8 27B measured
SAGE (Vals AI)42.5%52.4%Qwen3.8 27B
MedScribe (Vals AI)70.6%83.8%Qwen3.8 27B
Long context reasoning (AA-LCR)Not measured82.0%Only Qwen3.8 27B measured
DTBench82.9%88.0%Qwen3.8 27B
LMCA29.1%41.4%Qwen3.8 27B
Surface Evolver BenchNot measured45.0%Only Qwen3.8 27B measured
LMArena Text13971440Qwen3.8 27B
LMArena Text: instruction following13751436Qwen3.8 27B
LMArena Text: creative writing13431383Qwen3.8 27B
LMArena VisionNot measured1269Only Qwen3.8 27B measured
LMArena Text: English14041448Qwen3.8 27B
LMArena Text: non-English13851428Qwen3.8 27B
LMArena Text: Chinese14461501Qwen3.8 27B
LMArena Text: German13911442Qwen3.8 27B
LMArena Text: Japanese13691388Qwen3.8 27B
LMArena Text: Korean13441380Qwen3.8 27B
LMArena Text: French14131464Qwen3.8 27B
LMArena Text: Spanish13991448Qwen3.8 27B
LMArena Text: Russian13801414Qwen3.8 27B
LMArena Text: Polish13761447Qwen3.8 27B
LMArena Text: multi-turn13931440Qwen3.8 27B
LMArena Text: longer queries13921449Qwen3.8 27B
LMArena Text: writing and language13691405Qwen3.8 27B
LMArena Text: entertainment and media13471393Qwen3.8 27B
LMArena Vision: EnglishNot measured1267Only Qwen3.8 27B measured
LMArena Vision: ChineseNot measured1326Only Qwen3.8 27B measured
LMArena Vision: reading textNot measured1281Only Qwen3.8 27B measured
LMArena Vision: humorNot measured1235Only Qwen3.8 27B measured
LMArena Vision: creative writingNot measured1237Only Qwen3.8 27B measured
Neither model has a result on the other 18 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Qwen3.5-Flash and Qwen3.8 27B
BenchmarkQwen3.5-FlashQwen3.8 27BLeader
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

Qwen3.5-Flash$0.0002Fits in one request
Qwen3.8 27B$0.0017Fits in one request

Qwen3.5-Flash costs less

Repository review

50K input + 3K output tokens

Qwen3.5-Flash$0.0040Fits in one request
Qwen3.8 27B$0.029Fits in one request

Qwen3.5-Flash costs less

Agent loop

200K input + 10K output tokens

Qwen3.5-Flash$0.016Fits in one request
Qwen3.8 27B$0.111Fits in one request

Qwen3.5-Flash 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 Qwen3.5-Flash and Qwen3.8 27B
SpecificationQwen3.5-FlashQwen3.8 27B
MakerQwenQwen
WeightsNo public weights listedOpen weights
ReleasedFebruary 25, 2026August 14, 2026
Context window1M1M
Input $/Mtok$0.065$0.425
Output $/Mtok$0.26$2.55
Coding index41.174.1
Independent measurements behind it24
Cost per solved task$0.40$1.26
SourceOpenRouter listingOpenRouter listing

Questions

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

Pick Qwen3.8 27B for coding: it scores 74.1 on the Vibe Coding Index against 41.1 for Qwen3.5-Flash. Pick Qwen3.5-Flash for output tokens 9.8x cheaper ($0.26 vs $2.55 per million) and a lower cost per solved task ($0.40 against $1.26).

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

Qwen3.5-Flash costs $0.065 per million input tokens and $0.26 per million output tokens; Qwen3.8 27B costs $0.425 and $2.55. Per solved SWE-bench Verified task, Qwen3.5-Flash costs $0.40 and Qwen3.8 27B $1.26.

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

Neither: both take 1M tokens.

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

45 benchmarks have a result for both models as of October 23, 2026. Qwen3.5-Flash's coding index rests on 2 independent benchmarks and Qwen3.8 27B's on 4. 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/qwen3-5-flash-02-23-vs-qwen3-8-27b" title="Qwen3.5-Flash vs Qwen3.8 27B, 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/qwen3-5-flash-02-23-vs-qwen3-8-27b">LLMBenchmark.ai, Qwen3.5-Flash vs Qwen3.8 27B</a></p>