Enter opens the highlighted result. Esc closes.

GPT-5.4 Mini vs Qwen3 235B A22B Thinking 2507

GPT-5.4 Mini leads Qwen3 235B A22B Thinking 2507 on the Artificial Analysis Intelligence Index, 24.1 to 12.7, and Qwen3 235B A22B Thinking 2507 costs 1.3x less per output token ($3.50 vs $4.50 per million). Updated October 22, 2026; 37 benchmarks have a result for both.

Replace GPT-5.4 Mini Replace Qwen3 235B A22B Thinking 2507
GPT-5.4 MiniOpenAI
Coding
35.8
Overall
44.0
Output $/M
$4.50
23
Qwen3 235B A22B Thinking 2507Qwen
Coding
No data
Overall
52.9
Output $/M
$3.50
14
vs boards won of 37 shared

GPT-5.4 Mini or Qwen3 235B A22B Thinking 2507: which to pick

Only GPT-5.4 Mini has a coding score, 35.8 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 output tokens 1.3x cheaper ($3.50 vs $4.50 per million) and work measured by LMArena Text: expert or LMArena Text: hard prompts in English, where it leads.

Model AGPT-5.4 MiniOpenAI35.8/1005 independent benchmarks$0.70 per solved taskModel BQwen3 235B A22B Thinking 2507QwenNot benchmarkedNo coding benchmarkNo cost per solved task
Shared results
37
GPT-5.4 Mini only
52
Qwen3 235B A22B Thinking 2507 only
9
Like-for-like categories
7 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-5.4 Mini 35.8 from 5 benchmarks, Qwen3 235B A22B Thinking 2507 not benchmarked.
Widest category
Agentic work, 16.6 points apart on average over 1 shared benchmark.
Widest single gap
AA Coding Index: GPT-5.4 Mini 56.1 against Qwen3 235B A22B Thinking 2507 22.1, 34.0 points apart on a 0 to 100 scale.
Output price
GPT-5.4 Mini $4.50 against Qwen3 235B A22B Thinking 2507 $3.50 per million output tokens, 1.3x apart.
Context window
GPT-5.4 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-5.4 MiniBarsQwen3 235B A22B Thinking 2507Ties
Coding40
Intelligence10
Agentic work10
Reasoning32
Math21
Knowledge14
General117
All shared boards2314
CodingLike-for-likeSix of these boards feed the Vibe Coding Index40
Coding benchmarks for GPT-5.4 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5.4 MiniQwen3 235B A22B Thinking 2507Leader
LMArena WebDev1397Not measuredOnly GPT-5.4 Mini measured
AA Coding Index56.122.1GPT-5.4 Mini
LiveCodeBench (Vals AI)81.5%Not measuredOnly GPT-5.4 Mini measured
SWE-bench Verified (Vals AI)73.0%Not measuredOnly GPT-5.4 Mini measured
Terminal-Bench 4 (Vals AI)2.5%Not measuredOnly GPT-5.4 Mini measured
Vibe Code Bench (Vals AI)48.0%Not measuredOnly GPT-5.4 Mini measured
Code migration (Vals AI)12.9%Not measuredOnly GPT-5.4 Mini measured
ProgramBench (Vals AI)0.0%Not measuredOnly GPT-5.4 Mini measured
Terminal-Bench 2.1 (Vals AI)54.7%Not measuredOnly GPT-5.4 Mini measured
SciCode52.1%Not measuredOnly GPT-5.4 Mini measured
Terminal-Bench Hard (Artificial Analysis run)52.3%Not measuredOnly GPT-5.4 Mini measured
FrontierCode27.0%Not measuredOnly GPT-5.4 Mini measured
WeirdML60.3%41.0%GPT-5.4 Mini
ALE-Bench1188.6Not measuredOnly GPT-5.4 Mini measured
LMArena Text: coding14381424GPT-5.4 Mini
LMArena Text: software and IT14361431GPT-5.4 Mini
LMArena WebDev: React1386Not measuredOnly GPT-5.4 Mini measured
LMArena WebDev: HTML1419Not measuredOnly GPT-5.4 Mini measured
Neither model has a result on the other 21 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
Intelligence benchmarks for GPT-5.4 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5.4 MiniQwen3 235B A22B Thinking 2507Leader
AA Intelligence Index24.112.7GPT-5.4 Mini
Epoch Capabilities Index148.7Not measuredOnly GPT-5.4 Mini measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
Agentic work benchmarks for GPT-5.4 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5.4 MiniQwen3 235B A22B Thinking 2507Leader
AA Agentic Index17.91.3GPT-5.4 Mini
Finance Agent (Vals AI)45.4%Not measuredOnly GPT-5.4 Mini measured
Tax Agent Bench (Vals AI)8.8%Not measuredOnly GPT-5.4 Mini measured
Legal Research Bench (Vals AI)12.5%Not measuredOnly GPT-5.4 Mini measured
Excel Modeling Benchmark (Vals AI)45.4%Not measuredOnly GPT-5.4 Mini measured
τ²-Bench Telecom83.3%Not measuredOnly GPT-5.4 Mini measured
DeepResearch Bench36.3%Not measuredOnly GPT-5.4 Mini measured
Vending-Bench 2Not measured-$11Only Qwen3 235B A22B Thinking 2507 measured
Neither model has a result on the other 33 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index32
Reasoning benchmarks for GPT-5.4 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5.4 MiniQwen3 235B A22B Thinking 2507Leader
GPQA Diamond83.1%Not measuredOnly GPT-5.4 Mini measured
MMMU Pro (Vals AI)79.3%Not measuredOnly GPT-5.4 Mini measured
GPQA Diamond (Artificial Analysis run)87.5%Not measuredOnly GPT-5.4 Mini measured
Humanity's Last Exam (Artificial Analysis run)28.1%Not measuredOnly GPT-5.4 Mini measured
GPQA Diamond (Epoch AI run)86.9%80.1%GPT-5.4 Mini
ARC-AGI63.7%Not measuredOnly GPT-5.4 Mini measured
ARC-AGI-218.9%Not measuredOnly GPT-5.4 Mini measured
Chess puzzles23.0%Not measuredOnly GPT-5.4 Mini measured
Mystery game puzzles11.0%9.0%GPT-5.4 Mini
LMArena Text: hard prompts14241416GPT-5.4 Mini
LMArena Text: expert14351463Qwen3 235B A22B Thinking 2507
LMArena Text: hard prompts in English14201432Qwen3 235B A22B Thinking 2507
LMArena Vision: diagrams1266Not measuredOnly GPT-5.4 Mini measured
LMArena Vision: homework1285Not measuredOnly GPT-5.4 Mini measured
Kagi LLM Benchmark37.9%Not measuredOnly GPT-5.4 Mini measured
Neither model has a result on the other 7 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index21
Math benchmarks for GPT-5.4 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5.4 MiniQwen3 235B A22B Thinking 2507Leader
AIME95.6%Not measuredOnly GPT-5.4 Mini measured
OTIS Mock AIME88.9%86.7%GPT-5.4 Mini
FrontierMath, tiers 1 to 351.2%Not measuredOnly GPT-5.4 Mini measured
FrontierMath, tier 49.8%Not measuredOnly GPT-5.4 Mini measured
ProofBench21.0%Not measuredOnly GPT-5.4 Mini measured
LMArena Text: math14181412GPT-5.4 Mini
LMArena Text: mathematical work14291435Qwen3 235B A22B Thinking 2507
Neither model has a result on the other 6 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index14
Knowledge benchmarks for GPT-5.4 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5.4 MiniQwen3 235B A22B Thinking 2507Leader
MMLU Pro84.5%Not measuredOnly GPT-5.4 Mini measured
CaseLaw (Vals AI)51.7%Not measuredOnly GPT-5.4 Mini measured
CorpFin (Vals AI)60.9%Not measuredOnly GPT-5.4 Mini measured
TaxEval (Vals AI)71.2%Not measuredOnly GPT-5.4 Mini measured
MortgageTax (Vals AI)63.5%Not measuredOnly GPT-5.4 Mini measured
SimpleQA Verified29.4%40.4%Qwen3 235B A22B Thinking 2507
LMArena Text: science14171433Qwen3 235B A22B Thinking 2507
LMArena Text: medicine and healthcare13951456Qwen3 235B A22B Thinking 2507
LMArena Text: legal and government14141415Qwen3 235B A22B Thinking 2507
LMArena Text: business and finance14181413GPT-5.4 Mini
Neither model has a result on the other 8 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index6 boards, none shared
Web design benchmarks for GPT-5.4 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5.4 MiniQwen3 235B A22B Thinking 2507Leader
Design Arena: websitesNot measured1052Only Qwen3 235B A22B Thinking 2507 measured
Design Arena: UI componentsNot measured942Only Qwen3 235B A22B Thinking 2507 measured
Design Arena: game devNot measured956Only Qwen3 235B A22B Thinking 2507 measured
Design Arena: data vizNot measured950Only Qwen3 235B A22B Thinking 2507 measured
Design Arena: 3DNot measured997Only Qwen3 235B A22B Thinking 2507 measured
Design Arena: all code categoriesNot measured1032Only Qwen3 235B A22B Thinking 2507 measured
Neither model has a result on the other 3 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index117
General benchmarks for GPT-5.4 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5.4 MiniQwen3 235B A22B Thinking 2507Leader
SAGE (Vals AI)50.8%Not measuredOnly GPT-5.4 Mini measured
Vals Multimodal Index54.2%Not measuredOnly GPT-5.4 Mini measured
IFBench73.3%Not measuredOnly GPT-5.4 Mini measured
Long context reasoning (AA-LCR)77.0%Not measuredOnly GPT-5.4 Mini measured
DTBench80.0%80.3%Qwen3 235B A22B Thinking 2507
LMCA40.8%29.3%GPT-5.4 Mini
Fiction.LiveBenchNot measured75.0%Only Qwen3 235B A22B Thinking 2507 measured
Creative writing (Lech Mazur)Not measured82.4%Only Qwen3 235B A22B Thinking 2507 measured
LMArena Text14121415Qwen3 235B A22B Thinking 2507
LMArena Text: instruction following14051386GPT-5.4 Mini
LMArena Text: creative writing13701384Qwen3 235B A22B Thinking 2507
LMArena Vision1245Not measuredOnly GPT-5.4 Mini measured
LMArena Text: English14121429Qwen3 235B A22B Thinking 2507
LMArena Text: non-English14051397GPT-5.4 Mini
LMArena Text: Chinese14461481Qwen3 235B A22B Thinking 2507
LMArena Text: German14081408GPT-5.4 Mini
LMArena Text: Japanese13751386Qwen3 235B A22B Thinking 2507
LMArena Text: Korean13701375Qwen3 235B A22B Thinking 2507
LMArena Text: French1441Not measuredOnly GPT-5.4 Mini measured
LMArena Text: Spanish14051389GPT-5.4 Mini
LMArena Text: Russian14171401GPT-5.4 Mini
LMArena Text: Polish14151390GPT-5.4 Mini
LMArena Text: multi-turn14291406GPT-5.4 Mini
LMArena Text: longer queries14081400GPT-5.4 Mini
LMArena Text: writing and language13931388GPT-5.4 Mini
LMArena Text: entertainment and media13731370GPT-5.4 Mini
LMArena Vision: English1244Not measuredOnly GPT-5.4 Mini measured
LMArena Vision: Chinese1286Not measuredOnly GPT-5.4 Mini measured
LMArena Vision: reading text1253Not measuredOnly GPT-5.4 Mini measured
LMArena Vision: humor1235Not measuredOnly GPT-5.4 Mini measured
LMArena Vision: recognizing things1218Not measuredOnly GPT-5.4 Mini measured
LMArena Vision: creative writing1222Not measuredOnly GPT-5.4 Mini measured
Neither model has a result on the other 14 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for GPT-5.4 Mini and Qwen3 235B A22B Thinking 2507
BenchmarkGPT-5.4 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-5.4 Mini$0.0030Fits in one request
Qwen3 235B A22B Thinking 2507$0.0022Fits in one request

Qwen3 235B A22B Thinking 2507 costs less

Repository review

50K input + 3K output tokens

GPT-5.4 Mini$0.051Fits in one request
Qwen3 235B A22B Thinking 2507$0.033Fits in one request

Qwen3 235B A22B Thinking 2507 costs less

Agent loop

200K input + 10K output tokens

GPT-5.4 Mini$0.195Fits in one request
Qwen3 235B A22B Thinking 2507$0.125Exceeds the context window

Qwen3 235B A22B Thinking 2507 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-5.4 Mini and Qwen3 235B A22B Thinking 2507
SpecificationOpenAI GPT-5.4 MiniQwen3 235B A22B Thinking 2507
MakerOpenAIQwen
WeightsNo public weights listedOpen weights
ReleasedMarch 17, 2026July 25, 2025
Context window400K128K
Input $/Mtok$0.75$0.45
Output $/Mtok$4.50$3.50
Coding index35.8Not benchmarked
Independent measurements behind it50
Cost per solved task$0.70Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is GPT-5.4 Mini better than Qwen3 235B A22B Thinking 2507 for coding?

Only GPT-5.4 Mini has a coding score, 35.8 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 output tokens 1.3x cheaper ($3.50 vs $4.50 per million) and work measured by LMArena Text: expert or LMArena Text: hard prompts in English, where it leads.

Which is cheaper, GPT-5.4 Mini or Qwen3 235B A22B Thinking 2507?

GPT-5.4 Mini costs $0.75 per million input tokens and $4.50 per million output tokens; Qwen3 235B A22B Thinking 2507 costs $0.45 and $3.50. Cost per solved task is known for GPT-5.4 Mini only.

Which has the bigger context window, GPT-5.4 Mini or Qwen3 235B A22B Thinking 2507?

GPT-5.4 Mini, with 400K tokens against 128K.

How many benchmarks is the GPT-5.4 Mini vs Qwen3 235B A22B Thinking 2507 comparison based on?

37 benchmarks have a result for both models as of October 22, 2026. GPT-5.4 Mini's coding index rests on 5 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