LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

Qwen2.5 Coder 32B Instruct vs Qwen3 Coder 480B A35B

Updated October 21, 2026. 28 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Qwen2.5 Coder 32B Instruct Replace Qwen3 Coder 480B A35B
Qwen2.5 Coder 32B InstructQwen
Coding
30.5
Overall
6.7
Output $/M
$1.00
0
Qwen3 Coder 480B A35BQwen
Coding
50.0
Overall
11.9
Output $/M
$1.00
28
vs boards won of 28 shared

Decision reading

Pick Qwen3 Coder 480B A35B for coding: it scores 50.0 on the Vibe Coding Index against 30.5 for Qwen2.5 Coder 32B Instruct.

Qwen2.5 Coder 32B Instruct is neither cheaper nor larger in context, and leads no benchmark both models were measured on.

Model AQwen2.5 Coder 32B InstructQwen30.5/1002 independent benchmarksNo cost per solved taskModel BQwen3 Coder 480B A35BQwen50.0/1002 independent benchmarksNo cost per solved task
Shared results
28
Qwen2.5 Coder 32B Instruct only
7
Qwen3 Coder 480B A35B only
24
Like-for-like categories
6 of 9

Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.

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
Qwen2.5 Coder 32B Instruct 30.5 from 2 benchmarks, Qwen3 Coder 480B A35B 50.0 from 2 benchmarks.
Widest category
Coding, 24.1 points apart on average over 4 shared benchmarks.
Widest single gap
LiveCodeBench (Artificial Analysis run): Qwen2.5 Coder 32B Instruct 29.5% against Qwen3 Coder 480B A35B 58.5%, 29.0 points apart on a 0 to 100 scale.
Output price
Both cost $1.00 per million output tokens.
Context window
Qwen2.5 Coder 32B Instruct takes 33K tokens, Qwen3 Coder 480B A35B 262K.

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
CategoryQwen2.5 Coder 32B InstructBarsQwen3 Coder 480B A35BTies
Coding04
Intelligence01
Reasoning05
Math02
Knowledge05
General011
All shared boards028
CodingLike-for-likeFour of these boards feed the Vibe Coding Index04
Coding benchmarks for Qwen2.5 Coder 32B Instruct and Qwen3 Coder 480B A35B
BenchmarkQwen2.5 Coder 32B InstructQwen3 Coder 480B A35BLeader
Aider polyglot16.4%Not measuredOnly Qwen2.5 Coder 32B Instruct measured
SWE-bench Verified47.0%69.6%Qwen3 Coder 480B A35B
LMArena WebDevNot measured1275Only Qwen3 Coder 480B A35B measured
LiveCodeBench (Artificial Analysis run)29.5%58.5%Qwen3 Coder 480B A35B
Terminal-Bench Hard (Artificial Analysis run)Not measured18.9%Only Qwen3 Coder 480B A35B measured
Terminal-BenchNot measured27.2%Only Qwen3 Coder 480B A35B measured
GSO-BenchNot measured4.9%Only Qwen3 Coder 480B A35B measured
WeirdMLNot measured41.2%Only Qwen3 Coder 480B A35B measured
ALE-BenchNot measured461.4Only Qwen3 Coder 480B A35B measured
LMArena Text: coding12761412Qwen3 Coder 480B A35B
LMArena Text: software and IT12611393Qwen3 Coder 480B A35B
LMArena WebDev: ReactNot measured1257Only Qwen3 Coder 480B A35B measured
LMArena WebDev: HTMLNot measured1287Only Qwen3 Coder 480B A35B measured
Neither model has a result on the other 25 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
Intelligence benchmarks for Qwen2.5 Coder 32B Instruct and Qwen3 Coder 480B A35B
BenchmarkQwen2.5 Coder 32B InstructQwen3 Coder 480B A35BLeader
AA Intelligence Index6.711.9Qwen3 Coder 480B A35B
Epoch Capabilities Index119.5Not measuredOnly Qwen2.5 Coder 32B Instruct measured
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index1 board, none shared
Agentic work benchmarks for Qwen2.5 Coder 32B Instruct and Qwen3 Coder 480B A35B
BenchmarkQwen2.5 Coder 32B InstructQwen3 Coder 480B A35BLeader
τ²-Bench TelecomNot measured43.6%Only Qwen3 Coder 480B A35B measured
Neither model has a result on the other 38 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index05
Reasoning benchmarks for Qwen2.5 Coder 32B Instruct and Qwen3 Coder 480B A35B
BenchmarkQwen2.5 Coder 32B InstructQwen3 Coder 480B A35BLeader
GPQA Diamond (Artificial Analysis run)41.7%61.8%Qwen3 Coder 480B A35B
Humanity's Last Exam (Artificial Analysis run)3.5%4.5%Qwen3 Coder 480B A35B
LMArena Text: hard prompts12511372Qwen3 Coder 480B A35B
LMArena Text: expert12211338Qwen3 Coder 480B A35B
LMArena Text: hard prompts in English12691380Qwen3 Coder 480B A35B
Neither model has a result on the other 16 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for Qwen2.5 Coder 32B Instruct and Qwen3 Coder 480B A35B
BenchmarkQwen2.5 Coder 32B InstructQwen3 Coder 480B A35BLeader
AA Math IndexNot measured39.3Only Qwen3 Coder 480B A35B measured
AIME 2025 (Artificial Analysis run)Not measured39.3%Only Qwen3 Coder 480B A35B measured
GSM8K93.0%Not measuredOnly Qwen2.5 Coder 32B Instruct measured
LMArena Text: math12511365Qwen3 Coder 480B A35B
LMArena Text: mathematical work12431367Qwen3 Coder 480B A35B
Neither model has a result on the other 8 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index05
Knowledge benchmarks for Qwen2.5 Coder 32B Instruct and Qwen3 Coder 480B A35B
BenchmarkQwen2.5 Coder 32B InstructQwen3 Coder 480B A35BLeader
MMLU Pro (Artificial Analysis run)63.5%78.8%Qwen3 Coder 480B A35B
MMLU79.1%Not measuredOnly Qwen2.5 Coder 32B Instruct measured
AI2 Reasoning Challenge70.5%Not measuredOnly Qwen2.5 Coder 32B Instruct measured
LMArena Text: science12211347Qwen3 Coder 480B A35B
LMArena Text: medicine and healthcare11871347Qwen3 Coder 480B A35B
LMArena Text: legal and government12571358Qwen3 Coder 480B A35B
LMArena Text: business and finance12241343Qwen3 Coder 480B A35B
Neither model has a result on the other 10 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index5 boards, none shared
Web design benchmarks for Qwen2.5 Coder 32B Instruct and Qwen3 Coder 480B A35B
BenchmarkQwen2.5 Coder 32B InstructQwen3 Coder 480B A35BLeader
Design Arena: websitesNot measured1159Only Qwen3 Coder 480B A35B measured
Design Arena: UI componentsNot measured1112Only Qwen3 Coder 480B A35B measured
Design Arena: game devNot measured1099Only Qwen3 Coder 480B A35B measured
Design Arena: data vizNot measured1084Only Qwen3 Coder 480B A35B measured
Design Arena: all code categoriesNot measured1143Only Qwen3 Coder 480B A35B measured
Neither model has a result on the other 4 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index011
General benchmarks for Qwen2.5 Coder 32B Instruct and Qwen3 Coder 480B A35B
BenchmarkQwen2.5 Coder 32B InstructQwen3 Coder 480B A35BLeader
IFBenchNot measured40.5%Only Qwen3 Coder 480B A35B measured
Long context reasoning (AA-LCR)Not measured45.7%Only Qwen3 Coder 480B A35B measured
HellaSwag83.0%Not measuredOnly Qwen2.5 Coder 32B Instruct measured
WinoGrande80.8%Not measuredOnly Qwen2.5 Coder 32B Instruct measured
LMArena Text12301357Qwen3 Coder 480B A35B
LMArena Text: instruction following12231355Qwen3 Coder 480B A35B
LMArena Text: creative writing11741333Qwen3 Coder 480B A35B
LMArena Text: English12481359Qwen3 Coder 480B A35B
LMArena Text: non-English12051346Qwen3 Coder 480B A35B
LMArena Text: Chinese12221357Qwen3 Coder 480B A35B
LMArena Text: GermanNot measured1325Only Qwen3 Coder 480B A35B measured
LMArena Text: JapaneseNot measured1311Only Qwen3 Coder 480B A35B measured
LMArena Text: KoreanNot measured1305Only Qwen3 Coder 480B A35B measured
LMArena Text: FrenchNot measured1398Only Qwen3 Coder 480B A35B measured
LMArena Text: SpanishNot measured1360Only Qwen3 Coder 480B A35B measured
LMArena Text: Russian12281366Qwen3 Coder 480B A35B
LMArena Text: PolishNot measured1344Only Qwen3 Coder 480B A35B measured
LMArena Text: multi-turn12221366Qwen3 Coder 480B A35B
LMArena Text: longer queries12511378Qwen3 Coder 480B A35B
LMArena Text: writing and language11911340Qwen3 Coder 480B A35B
LMArena Text: entertainment and media11501317Qwen3 Coder 480B A35B
Neither model has a result on the other 25 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Qwen2.5 Coder 32B Instruct and Qwen3 Coder 480B A35B
BenchmarkQwen2.5 Coder 32B InstructQwen3 Coder 480B A35BLeader
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

Qwen2.5 Coder 32B Instruct$0.0012Fits in one request
Qwen3 Coder 480B A35B$0.0008Fits in one request

Qwen3 Coder 480B A35B costs less

Repository review

50K input + 3K output tokens

Qwen2.5 Coder 32B Instruct$0.036Exceeds the context window
Qwen3 Coder 480B A35B$0.018Fits in one request

Qwen3 Coder 480B A35B costs less

Agent loop

200K input + 10K output tokens

Qwen2.5 Coder 32B Instruct$0.142Exceeds the context window
Qwen3 Coder 480B A35B$0.070Fits in one request

Qwen3 Coder 480B A35B 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 Qwen2.5 Coder 32B Instruct and Qwen3 Coder 480B A35B
SpecificationQwen2.5 Coder 32B InstructQwen3 Coder 480B A35B
MakerQwenQwen
WeightsOpen weightsOpen weights
ReleasedNovember 11, 2024July 23, 2025
Context window33K262K
Input $/Mtok$0.66$0.30
Output $/Mtok$1.00$1.00
Coding index30.550.0
Independent measurements behind it22
Cost per solved taskNot computableNot computable
SourceOpenRouter listingOpenRouter listing

Questions

Is Qwen2.5 Coder 32B Instruct better than Qwen3 Coder 480B A35B for coding?

Pick Qwen3 Coder 480B A35B for coding: it scores 50.0 on the Vibe Coding Index against 30.5 for Qwen2.5 Coder 32B Instruct. Qwen2.5 Coder 32B Instruct is neither cheaper nor larger in context, and leads no benchmark both models were measured on.

Which is cheaper, Qwen2.5 Coder 32B Instruct or Qwen3 Coder 480B A35B?

Qwen2.5 Coder 32B Instruct costs $0.66 per million input tokens and $1.00 per million output tokens; Qwen3 Coder 480B A35B costs $0.30 and $1.00. Cost per solved task is known for neither of them.

Which has the bigger context window, Qwen2.5 Coder 32B Instruct or Qwen3 Coder 480B A35B?

Qwen3 Coder 480B A35B, with 262K tokens against 33K.

How many benchmarks is the Qwen2.5 Coder 32B Instruct vs Qwen3 Coder 480B A35B comparison based on?

28 benchmarks have a result for both models as of October 21, 2026. Qwen2.5 Coder 32B Instruct's coding index rests on 2 independent benchmarks and Qwen3 Coder 480B A35B's on 2. 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/qwen-2-5-coder-32b-instruct-vs-qwen3-coder" title="Qwen2.5 Coder 32B Instruct vs Qwen3 Coder 480B A35B, 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/qwen-2-5-coder-32b-instruct-vs-qwen3-coder">LLMBenchmark.ai, Qwen2.5 Coder 32B Instruct vs Qwen3 Coder 480B A35B</a></p>