Enter opens the highlighted result. Esc closes.

Claude Opus 4.8 vs Qwen3.8 27B

Qwen3.8 27B leads Claude Opus 4.8 on coding, 74.2 to 65.0 on the Vibe Coding Index, and Qwen3.8 27B costs 9.8x less per output token ($2.55 vs $25.00 per million). Updated October 22, 2026; 82 benchmarks have a result for both.

Replace Claude Opus 4.8 Replace Qwen3.8 27B
Claude Opus 4.8Anthropic
Coding
65.0
Overall
69.8
Output $/M
$25.00
74
Qwen3.8 27BQwen
Coding
74.2
Overall
59.6
Output $/M
$2.55
8
vs boards won of 82 shared

Claude Opus 4.8 or Qwen3.8 27B: which to pick

Pick Qwen3.8 27B for coding: it scores 74.2 on the Vibe Coding Index against 65.0 for Claude Opus 4.8, though on 4 benchmarks against 6.

Pick Claude Opus 4.8 for work measured by AA Coding Index or LiveCodeBench (Vals AI), where it leads.

Model AClaude Opus 4.8Anthropic65.0/1006 independent benchmarks$2.17 per solved taskModel BQwen3.8 27BQwen74.2/1004 independent benchmarks$1.26 per solved task
Shared results
82
Claude Opus 4.8 only
48
Qwen3.8 27B only
3
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
Claude Opus 4.8 65.0 from 6 benchmarks, Qwen3.8 27B 74.2 from 4 benchmarks.
Widest category
Math, 21.6 points apart on average over 3 shared benchmarks.
Widest single gap
LiveBench browser: Claude Opus 4.8 96.0% against Qwen3.8 27B 32.0%, 64.0 points apart on a 0 to 100 scale.
Output price
Claude Opus 4.8 $25.00 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
CategoryClaude Opus 4.8BarsQwen3.8 27BTies
Coding114
Intelligence20
Agentic work82
Reasoning120
Math30
Knowledge90
Web design10
General282
All shared boards748
CodingLike-for-likeSix of these boards feed the Vibe Coding Index114
Coding benchmarks for Claude Opus 4.8 and Qwen3.8 27B
BenchmarkClaude Opus 4.8Qwen3.8 27BLeader
LiveBench coding77.3%86.8%Qwen3.8 27B
LMArena WebDev15561593Qwen3.8 27B
AA Coding Index74.368.1Claude Opus 4.8
LiveCodeBench (Vals AI)87.8%84.0%Claude Opus 4.8
SWE-bench Verified (Vals AI)88.6%86.0%Claude Opus 4.8
Terminal-Bench 4 (Vals AI)23.2%Not measuredOnly Claude Opus 4.8 measured
Vibe Code Bench (Vals AI)82.7%64.8%Claude Opus 4.8
Code migration (Vals AI)47.3%14.2%Claude Opus 4.8
ProgramBench (Vals AI)1.0%0.0%Claude Opus 4.8
IOI (Vals AI)Not measured39.1%Only Qwen3.8 27B measured
Terminal-Bench 2.1 (Vals AI)71.9%58.4%Claude Opus 4.8
SkillsBench (Vals AI)59.2%38.1%Claude Opus 4.8
CyberBench (Vals AI)Not measured67.8%Only Qwen3.8 27B measured
SciCode54.4%46.6%Claude Opus 4.8
Terminal-Bench Hard (Artificial Analysis run)58.3%Not measuredOnly Claude Opus 4.8 measured
FrontierCode46.5%Not measuredOnly Claude Opus 4.8 measured
GSO-Bench47.1%Not measuredOnly Claude Opus 4.8 measured
WeirdML82.9%Not measuredOnly Claude Opus 4.8 measured
LMArena Text: coding14901483Claude Opus 4.8
LMArena Text: software and IT14881478Claude Opus 4.8
LMArena WebDev: React15571602Qwen3.8 27B
LMArena WebDev: HTML15431570Qwen3.8 27B
LMArena WebDev: from an imageNot measured1577Only Qwen3.8 27B measured
SciCode12.3%Not measuredOnly Claude Opus 4.8 measured
Neither model has a result on the other 15 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for Claude Opus 4.8 and Qwen3.8 27B
BenchmarkClaude Opus 4.8Qwen3.8 27BLeader
AA Intelligence Index41.833.7Claude Opus 4.8
Epoch Capabilities Index159.0151.7Claude Opus 4.8
Agentic workLike-for-likePublic boards, shown as published, kept out of the index82
Agentic work benchmarks for Claude Opus 4.8 and Qwen3.8 27B
BenchmarkClaude Opus 4.8Qwen3.8 27BLeader
AA Agentic Index41.945.8Qwen3.8 27B
Design Arena: agents building web apps1233Not measuredOnly Claude Opus 4.8 measured
Design Arena: agents building full-stack apps1249Not measuredOnly Claude Opus 4.8 measured
Design Arena: agents building mobile apps1222Not measuredOnly Claude Opus 4.8 measured
Design Arena: agents building Android apps1233Not measuredOnly Claude Opus 4.8 measured
Design Arena: agents building Godot games1231Not measuredOnly Claude Opus 4.8 measured
Design Arena: agents building games1239Not measuredOnly Claude Opus 4.8 measured
Design Arena: HTML slides1171Not measuredOnly Claude Opus 4.8 measured
Design Arena: PowerPoint slides1202Not measuredOnly Claude Opus 4.8 measured
Design Arena: spreadsheets1303Not measuredOnly Claude Opus 4.8 measured
LiveBench research81.8%59.4%Claude Opus 4.8
LiveBench browser96.0%32.0%Claude Opus 4.8
LiveBench long agentic48.3%27.9%Claude Opus 4.8
Finance Agent (Vals AI)53.9%48.5%Claude Opus 4.8
Tax Agent Bench (Vals AI)30.0%30.9%Qwen3.8 27B
Legal Research Bench (Vals AI)43.8%36.1%Claude Opus 4.8
Excel Modeling Benchmark (Vals AI)69.4%59.7%Claude Opus 4.8
Terminal-Bench-Science (Vals AI)4.3%0.0%Claude Opus 4.8
Time Horizon Index (Vals AI)11.8%Not measuredOnly Claude Opus 4.8 measured
τ²-Bench Telecom94.4%Not measuredOnly Claude Opus 4.8 measured
APEX-Agents48.9%47.5%Claude Opus 4.8
Remote Labor Index8.3%Not measuredOnly Claude Opus 4.8 measured
OSWorld 2.020.6%Not measuredOnly Claude Opus 4.8 measured
DeepResearch Bench50.2%Not measuredOnly Claude Opus 4.8 measured
PostTrainBench33.8%Not measuredOnly Claude Opus 4.8 measured
Vending-Bench 2$5,787Not measuredOnly Claude Opus 4.8 measured
LMArena Search1204Not measuredOnly Claude Opus 4.8 measured
Neither model has a result on the other 14 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index120
Reasoning benchmarks for Claude Opus 4.8 and Qwen3.8 27B
BenchmarkClaude Opus 4.8Qwen3.8 27BLeader
GPQA Diamond92.4%88.9%Claude Opus 4.8
LiveBench reasoning89.2%80.0%Claude Opus 4.8
MMMU Pro (Vals AI)86.6%83.9%Claude Opus 4.8
GPQA Diamond (Artificial Analysis run)92.0%90.5%Claude Opus 4.8
Humanity's Last Exam (Artificial Analysis run)48.7%33.9%Claude Opus 4.8
GPQA Diamond (Epoch AI run)91.0%Not measuredOnly Claude Opus 4.8 measured
ARC-AGI92.5%87.5%Claude Opus 4.8
ARC-AGI-272.1%42.4%Claude Opus 4.8
SimpleBench64.8%Not measuredOnly Claude Opus 4.8 measured
Chess puzzles52.0%Not measuredOnly Claude Opus 4.8 measured
Mystery game puzzles36.0%Not measuredOnly Claude Opus 4.8 measured
EBR-bench28.6%Not measuredOnly Claude Opus 4.8 measured
LMArena Text: hard prompts14821459Claude Opus 4.8
LMArena Text: expert15031479Claude Opus 4.8
LMArena Text: hard prompts in English14871461Claude Opus 4.8
LMArena Vision: diagrams13151285Claude Opus 4.8
LMArena Vision: homework13371295Claude Opus 4.8
Kagi LLM Benchmark88.8%Not measuredOnly Claude Opus 4.8 measured
Neither model has a result on the other 4 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index30
Math benchmarks for Claude Opus 4.8 and Qwen3.8 27B
BenchmarkClaude Opus 4.8Qwen3.8 27BLeader
OTIS Mock AIME98.3%Not measuredOnly Claude Opus 4.8 measured
FrontierMath, tiers 1 to 380.0%Not measuredOnly Claude Opus 4.8 measured
FrontierMath, tier 456.1%Not measuredOnly Claude Opus 4.8 measured
ProofBench69.0%16.0%Claude Opus 4.8
LMArena Text: math14861455Claude Opus 4.8
LMArena Text: mathematical work14971458Claude Opus 4.8
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index90
Knowledge benchmarks for Claude Opus 4.8 and Qwen3.8 27B
BenchmarkClaude Opus 4.8Qwen3.8 27BLeader
MMLU Pro89.6%84.3%Claude Opus 4.8
LegalBench (Vals AI)83.6%82.4%Claude Opus 4.8
CorpFin (Vals AI)66.7%Not measuredOnly Claude Opus 4.8 measured
TaxEval (Vals AI)75.6%70.8%Claude Opus 4.8
MortgageTax (Vals AI)69.9%64.9%Claude Opus 4.8
MedCode (Vals AI)53.2%28.7%Claude Opus 4.8
SimpleQA Verified53.0%Not measuredOnly Claude Opus 4.8 measured
LMArena Text: science14771467Claude Opus 4.8
LMArena Text: medicine and healthcare14641437Claude Opus 4.8
LMArena Text: legal and government14771447Claude Opus 4.8
LMArena Text: business and finance14581438Claude Opus 4.8
FACTS Benchmark Suite53.5%Not measuredOnly Claude Opus 4.8 measured
Neither model has a result on the other 6 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index10
Web design benchmarks for Claude Opus 4.8 and Qwen3.8 27B
BenchmarkClaude Opus 4.8Qwen3.8 27BLeader
Design Arena: websites1258Not measuredOnly Claude Opus 4.8 measured
Design Arena: UI components1255Not measuredOnly Claude Opus 4.8 measured
Design Arena: game dev1264Not measuredOnly Claude Opus 4.8 measured
Design Arena: data viz1240Not measuredOnly Claude Opus 4.8 measured
Design Arena: 3D1222Not measuredOnly Claude Opus 4.8 measured
Design Arena: all code categories1255Not measuredOnly Claude Opus 4.8 measured
Design Arena: SVG1195Not measuredOnly Claude Opus 4.8 measured
Design Arena: ASCII art1277Not measuredOnly Claude Opus 4.8 measured
LiveBench design78.8%75.9%Claude Opus 4.8
GeneralLike-for-likePublic boards, shown as published, kept out of the index282
General benchmarks for Claude Opus 4.8 and Qwen3.8 27B
BenchmarkClaude Opus 4.8Qwen3.8 27BLeader
LiveBench data analysis66.0%76.6%Qwen3.8 27B
Public Benefits Bench (Vals AI)68.1%Not measuredOnly Claude Opus 4.8 measured
SAGE (Vals AI)54.8%52.4%Claude Opus 4.8
MedScribe (Vals AI)85.8%83.8%Claude Opus 4.8
Vals Multimodal Index70.9%Not measuredOnly Claude Opus 4.8 measured
IFBench62.2%Not measuredOnly Claude Opus 4.8 measured
Long context reasoning (AA-LCR)77.7%82.0%Qwen3.8 27B
DTBench94.9%88.0%Claude Opus 4.8
LMCA57.5%41.4%Claude Opus 4.8
Surface Evolver Bench87.5%45.0%Claude Opus 4.8
Furniture assembly42.5%Not measuredOnly Claude Opus 4.8 measured
LMArena Text14611440Claude Opus 4.8
LMArena Text: instruction following14761436Claude Opus 4.8
LMArena Text: creative writing14541383Claude Opus 4.8
LMArena Vision12941269Claude Opus 4.8
LMArena Text: English14681448Claude Opus 4.8
LMArena Text: non-English14501428Claude Opus 4.8
LMArena Text: Chinese15061501Claude Opus 4.8
LMArena Text: German14721443Claude Opus 4.8
LMArena Text: Japanese14421387Claude Opus 4.8
LMArena Text: Korean14311384Claude Opus 4.8
LMArena Text: French14811465Claude Opus 4.8
LMArena Text: Spanish14651449Claude Opus 4.8
LMArena Text: Russian14741415Claude Opus 4.8
LMArena Text: Polish14591439Claude Opus 4.8
LMArena Text: multi-turn14761440Claude Opus 4.8
LMArena Text: longer queries14831448Claude Opus 4.8
LMArena Text: writing and language14611405Claude Opus 4.8
LMArena Text: entertainment and media14401393Claude Opus 4.8
LMArena Vision: English12931267Claude Opus 4.8
LMArena Vision: Chinese13461326Claude Opus 4.8
LMArena Vision: reading text13031281Claude Opus 4.8
LMArena Vision: humor12821235Claude Opus 4.8
LMArena Vision: creative writing13011237Claude Opus 4.8
LMArena Document1475Not measuredOnly Claude Opus 4.8 measured
Neither model has a result on the other 11 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Claude Opus 4.8 and Qwen3.8 27B
BenchmarkClaude Opus 4.8Qwen3.8 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

Claude Opus 4.8$0.018Fits in one request
Qwen3.8 27B$0.0017Fits in one request

Qwen3.8 27B costs less

Repository review

50K input + 3K output tokens

Claude Opus 4.8$0.325Fits in one request
Qwen3.8 27B$0.029Fits in one request

Qwen3.8 27B costs less

Agent loop

200K input + 10K output tokens

Claude Opus 4.8$1.25Fits in one request
Qwen3.8 27B$0.111Fits in one request

Qwen3.8 27B 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 Claude Opus 4.8 and Qwen3.8 27B
SpecificationAnthropic Claude Opus 4.8Qwen3.8 27B
MakerAnthropicQwen
WeightsNo public weights listedOpen weights
ReleasedMay 27, 2026August 14, 2026
Context window1M1M
Input $/Mtok$5.00$0.425
Output $/Mtok$25.00$2.55
Coding index65.074.2
Independent measurements behind it64
Cost per solved task$2.17$1.26
SourceOpenRouter listingOpenRouter listing

Questions

Is Claude Opus 4.8 better than Qwen3.8 27B for coding?

Pick Qwen3.8 27B for coding: it scores 74.2 on the Vibe Coding Index against 65.0 for Claude Opus 4.8, though on 4 benchmarks against 6. Pick Claude Opus 4.8 for work measured by AA Coding Index or LiveCodeBench (Vals AI), where it leads.

Which is cheaper, Claude Opus 4.8 or Qwen3.8 27B?

Claude Opus 4.8 costs $5.00 per million input tokens and $25.00 per million output tokens; Qwen3.8 27B costs $0.425 and $2.55. Per solved SWE-bench Verified task, Claude Opus 4.8 costs $2.17 and Qwen3.8 27B $1.26.

Which has the bigger context window, Claude Opus 4.8 or Qwen3.8 27B?

Neither: both take 1M tokens.

How many benchmarks is the Claude Opus 4.8 vs Qwen3.8 27B comparison based on?

82 benchmarks have a result for both models as of October 22, 2026. Claude Opus 4.8's coding index rests on 6 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/claude-opus-4-8-vs-qwen3-8-27b" title="Claude Opus 4.8 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/claude-opus-4-8-vs-qwen3-8-27b">LLMBenchmark.ai, Claude Opus 4.8 vs Qwen3.8 27B</a></p>