Enter opens the highlighted result. Esc closes.

GPT-5 vs Qwen3.8 27B

Qwen3.8 27B leads GPT-5 on coding, 74.2 to 38.8 on the Vibe Coding Index, and Qwen3.8 27B costs 3.9x less per output token ($2.55 vs $10.00 per million). Updated October 22, 2026; 61 benchmarks have a result for both.

Replace GPT-5 Replace Qwen3.8 27B
GPT-5OpenAI
Coding
38.8
Overall
51.5
Output $/M
$10.00
9
Qwen3.8 27BQwen
Coding
74.2
Overall
59.6
Output $/M
$2.55
52
vs boards won of 61 shared

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

Pick Qwen3.8 27B for coding: it scores 74.2 on the Vibe Coding Index against 38.8 for GPT-5.

Pick GPT-5 for a lower cost per solved task ($0.39 against $1.26) and work measured by LiveCodeBench (Vals AI) or ProofBench, where it leads.

Model AGPT-5OpenAI38.8/1003 independent benchmarks$0.39 per solved taskModel BQwen3.8 27BQwen74.2/1004 independent benchmarks$1.26 per solved task
Shared results
61
GPT-5 only
52
Qwen3.8 27B only
24
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
GPT-5 38.8 from 3 benchmarks, Qwen3.8 27B 74.2 from 4 benchmarks.
Widest category
Coding, 20.1 points apart on average over 8 shared benchmarks.
Widest single gap
Vibe Code Bench (Vals AI): GPT-5 20.1% against Qwen3.8 27B 64.8%, 44.8 points apart on a 0 to 100 scale.
Output price
GPT-5 $10.00 against Qwen3.8 27B $2.55 per million output tokens, 3.9x apart.
Context window
GPT-5 takes 400K tokens, Qwen3.8 27B 1M.

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
CategoryGPT-5BarsQwen3.8 27BTies
Coding17
Intelligence02
Reasoning011
Math12
Knowledge54
General226
All shared boards952
CodingLike-for-likeSix of these boards feed the Vibe Coding Index17
Coding benchmarks for GPT-5 and Qwen3.8 27B
BenchmarkGPT-5Qwen3.8 27BLeader
Aider polyglot88.0%Not measuredOnly GPT-5 measured
SWE-bench Verified74.4%Not measuredOnly GPT-5 measured
LiveBench codingNot measured86.8%Only Qwen3.8 27B measured
LMArena WebDev14181593Qwen3.8 27B
AA Coding Index37.868.1Qwen3.8 27B
LiveCodeBench (Vals AI)85.9%84.0%GPT-5
SWE-bench Verified (Vals AI)69.0%86.0%Qwen3.8 27B
Vibe Code Bench (Vals AI)20.1%64.8%Qwen3.8 27B
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
LiveCodeBench (Artificial Analysis run)84.6%Not measuredOnly GPT-5 measured
SciCodeNot measured46.6%Only Qwen3.8 27B measured
Terminal-Bench Hard (Artificial Analysis run)37.9%Not measuredOnly GPT-5 measured
Terminal-Bench49.6%Not measuredOnly GPT-5 measured
GSO-Bench6.9%Not measuredOnly GPT-5 measured
WeirdML60.7%Not measuredOnly GPT-5 measured
ALE-Bench807.6Not measuredOnly GPT-5 measured
LMArena Text: coding14361483Qwen3.8 27B
LMArena Text: software and IT14201478Qwen3.8 27B
LMArena WebDev: ReactNot measured1602Only Qwen3.8 27B measured
LMArena WebDev: HTML14271570Qwen3.8 27B
LMArena WebDev: from an imageNot measured1577Only Qwen3.8 27B measured
SciCode0.0%Not measuredOnly GPT-5 measured
Neither model has a result on the other 12 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for GPT-5 and Qwen3.8 27B
BenchmarkGPT-5Qwen3.8 27BLeader
AA Intelligence Index23.033.7Qwen3.8 27B
Epoch Capabilities Index150.0151.7Qwen3.8 27B
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index17 boards, none shared
Agentic work benchmarks for GPT-5 and Qwen3.8 27B
BenchmarkGPT-5Qwen3.8 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
τ²-Bench Telecom86.5%Not measuredOnly GPT-5 measured
APEX-AgentsNot measured47.5%Only Qwen3.8 27B measured
METR time horizons69.6%Not measuredOnly GPT-5 measured
Remote Labor Index1.7%Not measuredOnly GPT-5 measured
GDPval34.8%Not measuredOnly GPT-5 measured
BALROG32.8%Not measuredOnly GPT-5 measured
DeepResearch Bench49.6%Not measuredOnly GPT-5 measured
BrowseComp-Plus80.0%Not measuredOnly GPT-5 measured
Neither model has a result on the other 24 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index011
Reasoning benchmarks for GPT-5 and Qwen3.8 27B
BenchmarkGPT-5Qwen3.8 27BLeader
GPQA Diamond85.6%88.9%Qwen3.8 27B
LiveBench reasoningNot measured80.0%Only Qwen3.8 27B measured
MMMU Pro (Vals AI)81.5%83.9%Qwen3.8 27B
GPQA Diamond (Artificial Analysis run)85.4%90.5%Qwen3.8 27B
Humanity's Last Exam (Artificial Analysis run)28.5%33.9%Qwen3.8 27B
GPQA Diamond (Epoch AI run)86.2%Not measuredOnly GPT-5 measured
ARC-AGI65.7%87.5%Qwen3.8 27B
ARC-AGI-29.9%42.4%Qwen3.8 27B
SimpleBench56.7%Not measuredOnly GPT-5 measured
Chess puzzles36.0%Not measuredOnly GPT-5 measured
Mystery game puzzles23.0%Not measuredOnly GPT-5 measured
EBR-bench12.7%Not measuredOnly GPT-5 measured
VPCT66.0%Not measuredOnly GPT-5 measured
LMArena Text: hard prompts14151459Qwen3.8 27B
LMArena Text: expert14191479Qwen3.8 27B
LMArena Text: hard prompts in English14191461Qwen3.8 27B
LMArena Vision: diagrams12521285Qwen3.8 27B
LMArena Vision: homework12781295Qwen3.8 27B
Kagi LLM Benchmark72.7%Not measuredOnly GPT-5 measured
Neither model has a result on the other 3 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index12
Math benchmarks for GPT-5 and Qwen3.8 27B
BenchmarkGPT-5Qwen3.8 27BLeader
AIME93.4%Not measuredOnly GPT-5 measured
MATH 500 (Vals AI)96.0%Not measuredOnly GPT-5 measured
MGSM (Vals AI)92.8%Not measuredOnly GPT-5 measured
AA Math Index94.3Not measuredOnly GPT-5 measured
AIME 2025 (Artificial Analysis run)94.3%Not measuredOnly GPT-5 measured
OTIS Mock AIME91.4%Not measuredOnly GPT-5 measured
FrontierMath, tiers 1 to 355.4%Not measuredOnly GPT-5 measured
FrontierMath, tier 422.0%Not measuredOnly GPT-5 measured
ProofBench18.0%16.0%GPT-5
MATH level 598.1%Not measuredOnly GPT-5 measured
LMArena Text: math14071455Qwen3.8 27B
LMArena Text: mathematical work14051458Qwen3.8 27B
Neither model has a result on the other 1 board here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index54
Knowledge benchmarks for GPT-5 and Qwen3.8 27B
BenchmarkGPT-5Qwen3.8 27BLeader
MMLU Pro86.5%84.3%GPT-5
LegalBench (Vals AI)86.0%82.4%GPT-5
CaseLaw (Vals AI)66.5%Not measuredOnly GPT-5 measured
CorpFin (Vals AI)61.1%Not measuredOnly GPT-5 measured
TaxEval (Vals AI)73.4%70.8%GPT-5
MortgageTax (Vals AI)65.5%64.9%GPT-5
MedQA (Vals AI)96.3%Not measuredOnly GPT-5 measured
MedCode (Vals AI)49.6%28.7%GPT-5
MMLU Pro (Artificial Analysis run)87.1%Not measuredOnly GPT-5 measured
SimpleQA Verified50.1%Not measuredOnly GPT-5 measured
GeoBench81.0%Not measuredOnly GPT-5 measured
LMArena Text: science14161467Qwen3.8 27B
LMArena Text: medicine and healthcare14131437Qwen3.8 27B
LMArena Text: legal and government14311447Qwen3.8 27B
LMArena Text: business and finance14001438Qwen3.8 27B
FACTS Benchmark Suite56.5%Not measuredOnly GPT-5 measured
Neither model has a result on the other 2 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index9 boards, none shared
Web design benchmarks for GPT-5 and Qwen3.8 27B
BenchmarkGPT-5Qwen3.8 27BLeader
Design Arena: websites1184Not measuredOnly GPT-5 measured
Design Arena: UI components1177Not measuredOnly GPT-5 measured
Design Arena: game dev1185Not measuredOnly GPT-5 measured
Design Arena: data viz1221Not measuredOnly GPT-5 measured
Design Arena: 3D1055Not measuredOnly GPT-5 measured
Design Arena: all code categories1167Not measuredOnly GPT-5 measured
Design Arena: SVG1188Not measuredOnly GPT-5 measured
Design Arena: ASCII art1141Not measuredOnly GPT-5 measured
LiveBench designNot measured75.9%Only Qwen3.8 27B measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index226
General benchmarks for GPT-5 and Qwen3.8 27B
BenchmarkGPT-5Qwen3.8 27BLeader
LiveBench data analysisNot measured76.6%Only Qwen3.8 27B measured
SAGE (Vals AI)43.7%52.4%Qwen3.8 27B
MedScribe (Vals AI)83.7%83.8%Qwen3.8 27B
IFBench73.1%Not measuredOnly GPT-5 measured
Long context reasoning (AA-LCR)78.2%82.0%Qwen3.8 27B
DTBench90.7%88.0%GPT-5
LMCA40.0%41.4%Qwen3.8 27B
Fiction.LiveBench97.2%Not measuredOnly GPT-5 measured
Creative writing (Lech Mazur)86.0%Not measuredOnly GPT-5 measured
Surface Evolver BenchNot measured45.0%Only Qwen3.8 27B measured
LMArena Text14061440Qwen3.8 27B
LMArena Text: instruction following13881436Qwen3.8 27B
LMArena Text: creative writing13651383Qwen3.8 27B
LMArena Vision12321269Qwen3.8 27B
LMArena Text: English14091448Qwen3.8 27B
LMArena Text: non-English13971428Qwen3.8 27B
LMArena Text: Chinese14221501Qwen3.8 27B
LMArena Text: German14161443Qwen3.8 27B
LMArena Text: Japanese14091387GPT-5
LMArena Text: Korean13601384Qwen3.8 27B
LMArena Text: French14091465Qwen3.8 27B
LMArena Text: Spanish13991449Qwen3.8 27B
LMArena Text: Russian14071415Qwen3.8 27B
LMArena Text: Polish14311439Qwen3.8 27B
LMArena Text: multi-turn14261440Qwen3.8 27B
LMArena Text: longer queries13991448Qwen3.8 27B
LMArena Text: writing and language13841405Qwen3.8 27B
LMArena Text: entertainment and media13791393Qwen3.8 27B
LMArena Vision: English12311267Qwen3.8 27B
LMArena Vision: Chinese12671326Qwen3.8 27B
LMArena Vision: reading text12461281Qwen3.8 27B
LMArena Vision: humor12191235Qwen3.8 27B
LMArena Vision: captioning1212Not measuredOnly GPT-5 measured
LMArena Vision: recognizing things1263Not measuredOnly GPT-5 measured
LMArena Vision: creative writing12101237Qwen3.8 27B
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 GPT-5 and Qwen3.8 27B
BenchmarkGPT-5Qwen3.8 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

GPT-5$0.0063Fits in one request
Qwen3.8 27B$0.0017Fits in one request

Qwen3.8 27B costs less

Repository review

50K input + 3K output tokens

GPT-5$0.092Fits in one request
Qwen3.8 27B$0.029Fits in one request

Qwen3.8 27B costs less

Agent loop

200K input + 10K output tokens

GPT-5$0.350Fits 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 GPT-5 and Qwen3.8 27B
SpecificationOpenAI GPT-5Qwen3.8 27B
MakerOpenAIQwen
WeightsNo public weights listedOpen weights
ReleasedAugust 7, 2025August 14, 2026
Context window400K1M
Input $/Mtok$1.25$0.425
Output $/Mtok$10.00$2.55
Coding index38.874.2
Independent measurements behind it34
Cost per solved task$0.39$1.26
SourceOpenRouter listingOpenRouter listing

Questions

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

Pick Qwen3.8 27B for coding: it scores 74.2 on the Vibe Coding Index against 38.8 for GPT-5. Pick GPT-5 for a lower cost per solved task ($0.39 against $1.26) and work measured by LiveCodeBench (Vals AI) or ProofBench, where it leads.

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

GPT-5 costs $1.25 per million input tokens and $10.00 per million output tokens; Qwen3.8 27B costs $0.425 and $2.55. Per solved SWE-bench Verified task, GPT-5 costs $0.39 and Qwen3.8 27B $1.26.

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

Qwen3.8 27B, with 1M tokens against 400K.

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

61 benchmarks have a result for both models as of October 22, 2026. GPT-5's coding index rests on 3 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/gpt-5-vs-qwen3-8-27b" title="GPT-5 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/gpt-5-vs-qwen3-8-27b">LLMBenchmark.ai, GPT-5 vs Qwen3.8 27B</a></p>