Enter opens the highlighted result. Esc closes.

GPT-5 Mini vs Qwen3 235B A22B Thinking 2507

GPT-5 Mini leads Qwen3 235B A22B Thinking 2507 on the Artificial Analysis Intelligence Index, 20.6 to 12.7, and GPT-5 Mini costs 1.8x less per output token ($2.00 vs $3.50 per million). Updated October 22, 2026; 44 benchmarks have a result for both.

Replace GPT-5 Mini Replace Qwen3 235B A22B Thinking 2507
GPT-5 MiniOpenAI
Coding
34.2
Overall
43.7
Output $/M
$2.00
12
Qwen3 235B A22B Thinking 2507Qwen
Coding
No data
Overall
52.9
Output $/M
$3.50
31
vs boards won of 44 shared, 1 tied

GPT-5 Mini or Qwen3 235B A22B Thinking 2507: which to pick

Only GPT-5 Mini has a coding score, 34.2 on the Vibe Coding Index; Qwen3 235B A22B Thinking 2507 has none, so the two cannot be ranked on coding.

Pick Qwen3 235B A22B Thinking 2507 for work measured by LMArena Text: coding or LMArena Text: software and IT, where it leads.

Model AGPT-5 MiniOpenAI34.2/1002 independent benchmarks$0.08 per solved taskModel BQwen3 235B A22B Thinking 2507QwenNot benchmarkedNo coding benchmarkNo cost per solved task
Shared results
44
GPT-5 Mini only
57
Qwen3 235B A22B Thinking 2507 only
2
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
GPT-5 Mini 34.2 from 2 benchmarks, Qwen3 235B A22B Thinking 2507 not benchmarked.
Widest category
Web design, 14.8 points apart on average over 6 shared benchmarks.
Widest single gap
Design Arena: data viz: GPT-5 Mini 1131 against Qwen3 235B A22B Thinking 2507 950, 22.6 points apart on a 0 to 100 scale.
Output price
GPT-5 Mini $2.00 against Qwen3 235B A22B Thinking 2507 $3.50 per million output tokens, 1.8x apart.
Context window
GPT-5 Mini takes 400K tokens, Qwen3 235B A22B Thinking 2507 128K.

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-5 MiniBarsQwen3 235B A22B Thinking 2507Ties
Coding12
Intelligence10
Agentic work01
Reasoning14
Math021 tie
Knowledge05
Web design60
General317
All shared boards12311 tie
CodingLike-for-likeSix of these boards feed the Vibe Coding Index12
Coding benchmarks for GPT-5 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5 MiniQwen3 235B A22B Thinking 2507Leader
SWE-bench Verified59.8%Not measuredOnly GPT-5 Mini measured
SWE-bench Multilingual39.7%Not measuredOnly GPT-5 Mini measured
AA Coding IndexNot measured22.1Only Qwen3 235B A22B Thinking 2507 measured
LiveCodeBench (Vals AI)86.6%Not measuredOnly GPT-5 Mini measured
SWE-bench Verified (Vals AI)60.8%Not measuredOnly GPT-5 Mini measured
Vibe Code Bench (Vals AI)14.2%Not measuredOnly GPT-5 Mini measured
LiveCodeBench (Artificial Analysis run)83.8%Not measuredOnly GPT-5 Mini measured
SciCode39.0%Not measuredOnly GPT-5 Mini measured
Terminal-Bench Hard (Artificial Analysis run)33.3%Not measuredOnly GPT-5 Mini measured
Terminal-Bench34.8%Not measuredOnly GPT-5 Mini measured
WeirdML52.7%41.0%GPT-5 Mini
LMArena Text: coding14061424Qwen3 235B A22B Thinking 2507
LMArena Text: software and IT13941431Qwen3 235B A22B Thinking 2507
Neither model has a result on the other 26 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
Intelligence benchmarks for GPT-5 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5 MiniQwen3 235B A22B Thinking 2507Leader
AA Intelligence Index20.612.7GPT-5 Mini
Epoch Capabilities Index145.3Not measuredOnly GPT-5 Mini measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
Agentic work benchmarks for GPT-5 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5 MiniQwen3 235B A22B Thinking 2507Leader
AA Agentic IndexNot measured1.3Only Qwen3 235B A22B Thinking 2507 measured
τ²-Bench Telecom71.1%Not measuredOnly GPT-5 Mini measured
Vending-Bench 2-$31-$11Qwen3 235B A22B Thinking 2507
Berkeley Function Calling Leaderboard (BFCL)55.5%Not measuredOnly GPT-5 Mini measured
Neither model has a result on the other 37 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index14
Reasoning benchmarks for GPT-5 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5 MiniQwen3 235B A22B Thinking 2507Leader
GPQA Diamond80.3%Not measuredOnly GPT-5 Mini measured
MMMU Pro (Vals AI)78.9%Not measuredOnly GPT-5 Mini measured
GPQA Diamond (Artificial Analysis run)82.8%Not measuredOnly GPT-5 Mini measured
Humanity's Last Exam (Artificial Analysis run)21.5%Not measuredOnly GPT-5 Mini measured
GPQA Diamond (Epoch AI run)75.0%80.1%Qwen3 235B A22B Thinking 2507
ARC-AGI54.3%Not measuredOnly GPT-5 Mini measured
ARC-AGI-24.4%Not measuredOnly GPT-5 Mini measured
Chess puzzles24.0%Not measuredOnly GPT-5 Mini measured
Mystery game puzzles10.0%9.0%GPT-5 Mini
VPCT40.2%Not measuredOnly GPT-5 Mini measured
LMArena Text: hard prompts13801416Qwen3 235B A22B Thinking 2507
LMArena Text: expert13791463Qwen3 235B A22B Thinking 2507
LMArena Text: hard prompts in English13831432Qwen3 235B A22B Thinking 2507
LMArena Vision: diagrams1202Not measuredOnly GPT-5 Mini measured
LMArena Vision: homework1210Not measuredOnly GPT-5 Mini measured
Kagi LLM Benchmark70.3%Not measuredOnly GPT-5 Mini measured
Neither model has a result on the other 6 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for GPT-5 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5 MiniQwen3 235B A22B Thinking 2507Leader
AIME91.5%Not measuredOnly GPT-5 Mini measured
MATH 500 (Vals AI)94.8%Not measuredOnly GPT-5 Mini measured
MGSM (Vals AI)92.6%Not measuredOnly GPT-5 Mini measured
AA Math Index90.7Not measuredOnly GPT-5 Mini measured
AIME 2025 (Artificial Analysis run)90.7%Not measuredOnly GPT-5 Mini measured
OTIS Mock AIME86.7%86.7%Tie
FrontierMath, tiers 1 to 346.7%Not measuredOnly GPT-5 Mini measured
FrontierMath, tier 412.2%Not measuredOnly GPT-5 Mini measured
ProofBench9.0%Not measuredOnly GPT-5 Mini measured
MATH level 597.8%Not measuredOnly GPT-5 Mini measured
LMArena Text: math13781412Qwen3 235B A22B Thinking 2507
LMArena Text: mathematical work13801435Qwen3 235B A22B Thinking 2507
Neither model has a result on the other 1 board here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index05
Knowledge benchmarks for GPT-5 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5 MiniQwen3 235B A22B Thinking 2507Leader
MMLU Pro82.2%Not measuredOnly GPT-5 Mini measured
LegalBench (Vals AI)81.8%Not measuredOnly GPT-5 Mini measured
CaseLaw (Vals AI)68.5%Not measuredOnly GPT-5 Mini measured
CorpFin (Vals AI)60.2%Not measuredOnly GPT-5 Mini measured
TaxEval (Vals AI)75.2%Not measuredOnly GPT-5 Mini measured
MortgageTax (Vals AI)66.9%Not measuredOnly GPT-5 Mini measured
MedQA (Vals AI)96.1%Not measuredOnly GPT-5 Mini measured
MedCode (Vals AI)43.0%Not measuredOnly GPT-5 Mini measured
MMLU Pro (Artificial Analysis run)83.7%Not measuredOnly GPT-5 Mini measured
SimpleQA Verified21.6%40.4%Qwen3 235B A22B Thinking 2507
LMArena Text: science13711433Qwen3 235B A22B Thinking 2507
LMArena Text: medicine and healthcare13801456Qwen3 235B A22B Thinking 2507
LMArena Text: legal and government13751415Qwen3 235B A22B Thinking 2507
LMArena Text: business and finance13581413Qwen3 235B A22B Thinking 2507
FACTS Benchmark Suite38.6%Not measuredOnly GPT-5 Mini measured
Neither model has a result on the other 3 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index60
Web design benchmarks for GPT-5 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5 MiniQwen3 235B A22B Thinking 2507Leader
Design Arena: websites11251052GPT-5 Mini
Design Arena: UI components1103942GPT-5 Mini
Design Arena: game dev1129956GPT-5 Mini
Design Arena: data viz1131950GPT-5 Mini
Design Arena: 3D1036997GPT-5 Mini
Design Arena: all code categories11141032GPT-5 Mini
Design Arena: SVG1091Not measuredOnly GPT-5 Mini measured
Design Arena: ASCII art1125Not measuredOnly GPT-5 Mini measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index317
General benchmarks for GPT-5 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5 MiniQwen3 235B A22B Thinking 2507Leader
SAGE (Vals AI)43.0%Not measuredOnly GPT-5 Mini measured
MedScribe (Vals AI)80.6%Not measuredOnly GPT-5 Mini measured
IFBench75.4%Not measuredOnly GPT-5 Mini measured
Long context reasoning (AA-LCR)72.3%Not measuredOnly GPT-5 Mini measured
DTBench80.5%80.3%GPT-5 Mini
LMCA34.2%29.3%GPT-5 Mini
Fiction.LiveBench69.4%75.0%Qwen3 235B A22B Thinking 2507
Creative writing (Lech Mazur)83.1%82.4%GPT-5 Mini
LMArena Text13731415Qwen3 235B A22B Thinking 2507
LMArena Text: instruction following13571386Qwen3 235B A22B Thinking 2507
LMArena Text: creative writing13251384Qwen3 235B A22B Thinking 2507
LMArena Vision1202Not measuredOnly GPT-5 Mini measured
LMArena Text: English13791429Qwen3 235B A22B Thinking 2507
LMArena Text: non-English13631397Qwen3 235B A22B Thinking 2507
LMArena Text: Chinese13861481Qwen3 235B A22B Thinking 2507
LMArena Text: German13661408Qwen3 235B A22B Thinking 2507
LMArena Text: Japanese13411386Qwen3 235B A22B Thinking 2507
LMArena Text: Korean13081375Qwen3 235B A22B Thinking 2507
LMArena Text: French1386Not measuredOnly GPT-5 Mini measured
LMArena Text: Spanish13541389Qwen3 235B A22B Thinking 2507
LMArena Text: Russian13621401Qwen3 235B A22B Thinking 2507
LMArena Text: Polish13781390Qwen3 235B A22B Thinking 2507
LMArena Text: multi-turn13631406Qwen3 235B A22B Thinking 2507
LMArena Text: longer queries13551400Qwen3 235B A22B Thinking 2507
LMArena Text: writing and language13441388Qwen3 235B A22B Thinking 2507
LMArena Text: entertainment and media13401370Qwen3 235B A22B Thinking 2507
LMArena Vision: English1216Not measuredOnly GPT-5 Mini measured
LMArena Vision: Chinese1214Not measuredOnly GPT-5 Mini measured
LMArena Vision: reading text1199Not measuredOnly GPT-5 Mini measured
LMArena Vision: humor1205Not measuredOnly GPT-5 Mini measured
LMArena Vision: captioning1217Not measuredOnly GPT-5 Mini measured
LMArena Vision: recognizing things1226Not measuredOnly GPT-5 Mini measured
LMArena Vision: creative writing1181Not measuredOnly GPT-5 Mini measured
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 GPT-5 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5 MiniQwen3 235B A22B Thinking 2507Leader
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 Mini$0.0013Fits in one request
Qwen3 235B A22B Thinking 2507$0.0022Fits in one request

GPT-5 Mini costs less

Repository review

50K input + 3K output tokens

GPT-5 Mini$0.018Fits in one request
Qwen3 235B A22B Thinking 2507$0.033Fits in one request

GPT-5 Mini costs less

Agent loop

200K input + 10K output tokens

GPT-5 Mini$0.070Fits in one request
Qwen3 235B A22B Thinking 2507$0.125Exceeds the context window

GPT-5 Mini 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 Mini and Qwen3 235B A22B Thinking 2507
SpecificationOpenAI GPT-5 MiniQwen3 235B A22B Thinking 2507
MakerOpenAIQwen
WeightsNo public weights listedOpen weights
ReleasedAugust 7, 2025July 25, 2025
Context window400K128K
Input $/Mtok$0.25$0.45
Output $/Mtok$2.00$3.50
Coding index34.2Not benchmarked
Independent measurements behind it20
Cost per solved task$0.08Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is GPT-5 Mini better than Qwen3 235B A22B Thinking 2507 for coding?

Only GPT-5 Mini has a coding score, 34.2 on the Vibe Coding Index; Qwen3 235B A22B Thinking 2507 has none, so the two cannot be ranked on coding. Pick Qwen3 235B A22B Thinking 2507 for work measured by LMArena Text: coding or LMArena Text: software and IT, where it leads.

Which is cheaper, GPT-5 Mini or Qwen3 235B A22B Thinking 2507?

GPT-5 Mini costs $0.25 per million input tokens and $2.00 per million output tokens; Qwen3 235B A22B Thinking 2507 costs $0.45 and $3.50. Cost per solved task is known for GPT-5 Mini only.

Which has the bigger context window, GPT-5 Mini or Qwen3 235B A22B Thinking 2507?

GPT-5 Mini, with 400K tokens against 128K.

How many benchmarks is the GPT-5 Mini vs Qwen3 235B A22B Thinking 2507 comparison based on?

44 benchmarks have a result for both models as of October 22, 2026. GPT-5 Mini's coding index rests on 2 independent benchmarks and Qwen3 235B A22B Thinking 2507's on 0. Where only one model has a result, the page says so instead of naming a winner.

Compare next