Enter opens the highlighted result. Esc closes.

Llama 4 Scout vs Qwen3 VL 235B A22B Instruct

Qwen3 VL 235B A22B Instruct leads Llama 4 Scout on the Artificial Analysis Intelligence Index, 9.9 to 6.5, and Llama 4 Scout costs 3.8x less per output token ($0.50 vs $1.90 per million). Updated October 22, 2026; 47 benchmarks have a result for both.

Replace Llama 4 Scout Replace Qwen3 VL 235B A22B Instruct
Llama 4 ScoutMeta Llama
Coding
No data
Overall
24.9
Output $/M
$0.50
0
Qwen3 VL 235B A22B InstructQwen
Coding
No data
Overall
No data
Output $/M
$1.90
47
vs boards won of 47 shared

Llama 4 Scout or Qwen3 VL 235B A22B Instruct: which to pick

Neither model has a coding score yet, so the two cannot be ranked on coding.

Pick Llama 4 Scout for output tokens 3.8x cheaper ($0.50 vs $1.90 per million) and a larger context window (1.31M against 262K tokens).

Model ALlama 4 ScoutMeta LlamaNot benchmarkedNo coding benchmark$1.276 per solved taskModel BQwen3 VL 235B A22B InstructQwenNot benchmarkedNo coding benchmarkNo cost per solved task
Shared results
47
Llama 4 Scout only
35
Qwen3 VL 235B A22B Instruct only
2
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
Llama 4 Scout not benchmarked, Qwen3 VL 235B A22B Instruct not benchmarked.
Widest category
Math, 48.5 points apart on average over 4 shared benchmarks.
Widest single gap
AIME 2025 (Artificial Analysis run): Llama 4 Scout 14.0% against Qwen3 VL 235B A22B Instruct 88.3%, 74.3 points apart on a 0 to 100 scale.
Output price
Llama 4 Scout $0.50 against Qwen3 VL 235B A22B Instruct $1.90 per million output tokens, 3.8x apart.
Context window
Llama 4 Scout takes 1.31M tokens, Qwen3 VL 235B A22B Instruct 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
CategoryLlama 4 ScoutBarsQwen3 VL 235B A22B InstructTies
Coding04
Intelligence01
Agentic work01
Reasoning07
Math04
Knowledge05
General025
All shared boards047
CodingLike-for-likeSix of these boards feed the Vibe Coding Index04
Coding benchmarks for Llama 4 Scout and Qwen3 VL 235B A22B Instruct
BenchmarkLlama 4 ScoutQwen3 VL 235B A22B InstructLeader
SWE-bench Verified9.1%Not measuredOnly Llama 4 Scout measured
AA Coding Index8.2Not measuredOnly Llama 4 Scout measured
LiveCodeBench (Vals AI)38.5%Not measuredOnly Llama 4 Scout measured
LiveCodeBench (Artificial Analysis run)29.9%64.6%Qwen3 VL 235B A22B Instruct
SciCode21.3%Not measuredOnly Llama 4 Scout measured
Terminal-Bench Hard (Artificial Analysis run)1.5%11.4%Qwen3 VL 235B A22B Instruct
LMArena Text: coding12861438Qwen3 VL 235B A22B Instruct
LMArena Text: software and IT12861442Qwen3 VL 235B A22B Instruct
Neither model has a result on the other 35 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
Intelligence benchmarks for Llama 4 Scout and Qwen3 VL 235B A22B Instruct
BenchmarkLlama 4 ScoutQwen3 VL 235B A22B InstructLeader
AA Intelligence Index6.59.9Qwen3 VL 235B A22B Instruct
Epoch Capabilities Index129.6Not measuredOnly Llama 4 Scout measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
Agentic work benchmarks for Llama 4 Scout and Qwen3 VL 235B A22B Instruct
BenchmarkLlama 4 ScoutQwen3 VL 235B A22B InstructLeader
AA Agentic Index0.5Not measuredOnly Llama 4 Scout measured
τ²-Bench Telecom15.5%54.1%Qwen3 VL 235B A22B Instruct
Berkeley Function Calling Leaderboard (BFCL)28.1%Not measuredOnly Llama 4 Scout measured
Neither model has a result on the other 39 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index07
Reasoning benchmarks for Llama 4 Scout and Qwen3 VL 235B A22B Instruct
BenchmarkLlama 4 ScoutQwen3 VL 235B A22B InstructLeader
GPQA Diamond47.0%Not measuredOnly Llama 4 Scout measured
MMMU Pro (Vals AI)58.8%Not measuredOnly Llama 4 Scout measured
GPQA Diamond (Artificial Analysis run)58.7%77.2%Qwen3 VL 235B A22B Instruct
Humanity's Last Exam (Artificial Analysis run)3.8%11.9%Qwen3 VL 235B A22B Instruct
GPQA Diamond (Epoch AI run)51.8%Not measuredOnly Llama 4 Scout measured
ARC-AGI0.5%Not measuredOnly Llama 4 Scout measured
ARC-AGI-20.0%Not measuredOnly Llama 4 Scout measured
LMArena Text: hard prompts12661428Qwen3 VL 235B A22B Instruct
LMArena Text: expert12351442Qwen3 VL 235B A22B Instruct
LMArena Text: hard prompts in English12931439Qwen3 VL 235B A22B Instruct
LMArena Vision: diagrams11251255Qwen3 VL 235B A22B Instruct
LMArena Vision: homework11481277Qwen3 VL 235B A22B Instruct
Kagi LLM Benchmark36.9%Not measuredOnly Llama 4 Scout measured
Neither model has a result on the other 10 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index04
Math benchmarks for Llama 4 Scout and Qwen3 VL 235B A22B Instruct
BenchmarkLlama 4 ScoutQwen3 VL 235B A22B InstructLeader
AIME19.0%Not measuredOnly Llama 4 Scout measured
MATH 500 (Vals AI)79.2%Not measuredOnly Llama 4 Scout measured
MGSM (Vals AI)88.0%Not measuredOnly Llama 4 Scout measured
AA Math Index14.088.3Qwen3 VL 235B A22B Instruct
AIME 2025 (Artificial Analysis run)14.0%88.3%Qwen3 VL 235B A22B Instruct
OTIS Mock AIME7.8%Not measuredOnly Llama 4 Scout measured
MATH level 562.3%Not measuredOnly Llama 4 Scout measured
LMArena Text: math12871425Qwen3 VL 235B A22B Instruct
LMArena Text: mathematical work12861419Qwen3 VL 235B A22B Instruct
Neither model has a result on the other 5 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index05
Knowledge benchmarks for Llama 4 Scout and Qwen3 VL 235B A22B Instruct
BenchmarkLlama 4 ScoutQwen3 VL 235B A22B InstructLeader
MMLU Pro69.6%Not measuredOnly Llama 4 Scout measured
LegalBench (Vals AI)72.0%Not measuredOnly Llama 4 Scout measured
CorpFin (Vals AI)46.8%Not measuredOnly Llama 4 Scout measured
TaxEval (Vals AI)55.2%Not measuredOnly Llama 4 Scout measured
MortgageTax (Vals AI)57.8%Not measuredOnly Llama 4 Scout measured
MedQA (Vals AI)50.9%Not measuredOnly Llama 4 Scout measured
MedCode (Vals AI)23.3%Not measuredOnly Llama 4 Scout measured
MMLU Pro (Artificial Analysis run)75.2%83.6%Qwen3 VL 235B A22B Instruct
LMArena Text: science12781444Qwen3 VL 235B A22B Instruct
LMArena Text: medicine and healthcare12751431Qwen3 VL 235B A22B Instruct
LMArena Text: legal and government12851438Qwen3 VL 235B A22B Instruct
LMArena Text: business and finance12571433Qwen3 VL 235B A22B Instruct
Neither model has a result on the other 8 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index5 boards, none shared
Web design benchmarks for Llama 4 Scout and Qwen3 VL 235B A22B Instruct
BenchmarkLlama 4 ScoutQwen3 VL 235B A22B InstructLeader
Design Arena: websites750Not measuredOnly Llama 4 Scout measured
Design Arena: UI components765Not measuredOnly Llama 4 Scout measured
Design Arena: game dev772Not measuredOnly Llama 4 Scout measured
Design Arena: data viz896Not measuredOnly Llama 4 Scout measured
Design Arena: all code categories789Not measuredOnly Llama 4 Scout measured
Neither model has a result on the other 4 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index025
General benchmarks for Llama 4 Scout and Qwen3 VL 235B A22B Instruct
BenchmarkLlama 4 ScoutQwen3 VL 235B A22B InstructLeader
SAGE (Vals AI)34.8%Not measuredOnly Llama 4 Scout measured
MedScribe (Vals AI)50.6%Not measuredOnly Llama 4 Scout measured
IFBench39.5%56.5%Qwen3 VL 235B A22B Instruct
Long context reasoning (AA-LCR)27.7%63.7%Qwen3 VL 235B A22B Instruct
DTBench57.9%Not measuredOnly Llama 4 Scout measured
LMCA12.0%Not measuredOnly Llama 4 Scout measured
Fiction.LiveBench36.0%Not measuredOnly Llama 4 Scout measured
LMArena Text12791420Qwen3 VL 235B A22B Instruct
LMArena Text: instruction following12481406Qwen3 VL 235B A22B Instruct
LMArena Text: creative writing12491366Qwen3 VL 235B A22B Instruct
LMArena Vision11181247Qwen3 VL 235B A22B Instruct
LMArena Text: English13041429Qwen3 VL 235B A22B Instruct
LMArena Text: non-English12521405Qwen3 VL 235B A22B Instruct
LMArena Text: Chinese12551463Qwen3 VL 235B A22B Instruct
LMArena Text: German12731424Qwen3 VL 235B A22B Instruct
LMArena Text: Japanese12061385Qwen3 VL 235B A22B Instruct
LMArena Text: Korean12071394Qwen3 VL 235B A22B Instruct
LMArena Text: French12831452Qwen3 VL 235B A22B Instruct
LMArena Text: Spanish12781428Qwen3 VL 235B A22B Instruct
LMArena Text: Russian12631408Qwen3 VL 235B A22B Instruct
LMArena Text: Polish12651419Qwen3 VL 235B A22B Instruct
LMArena Text: multi-turn12801428Qwen3 VL 235B A22B Instruct
LMArena Text: longer queries12651419Qwen3 VL 235B A22B Instruct
LMArena Text: writing and language12691392Qwen3 VL 235B A22B Instruct
LMArena Text: entertainment and media12421368Qwen3 VL 235B A22B Instruct
LMArena Vision: English11431256Qwen3 VL 235B A22B Instruct
LMArena Vision: Chinese11061283Qwen3 VL 235B A22B Instruct
LMArena Vision: reading text11171253Qwen3 VL 235B A22B Instruct
LMArena Vision: humor11021209Qwen3 VL 235B A22B Instruct
LMArena Vision: captioningNot measured1281Only Qwen3 VL 235B A22B Instruct measured
LMArena Vision: recognizing thingsNot measured1249Only Qwen3 VL 235B A22B Instruct measured
LMArena Vision: creative writing11261220Qwen3 VL 235B A22B Instruct
Neither model has a result on the other 16 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Llama 4 Scout and Qwen3 VL 235B A22B Instruct
BenchmarkLlama 4 ScoutQwen3 VL 235B A22B InstructLeader
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

Llama 4 Scout$0.0004Fits in one request
Qwen3 VL 235B A22B Instruct$0.0012Fits in one request

Llama 4 Scout costs less

Repository review

50K input + 3K output tokens

Llama 4 Scout$0.010Fits in one request
Qwen3 VL 235B A22B Instruct$0.016Fits in one request

Llama 4 Scout costs less

Agent loop

200K input + 10K output tokens

Llama 4 Scout$0.040Fits in one request
Qwen3 VL 235B A22B Instruct$0.061Fits in one request

Llama 4 Scout 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 Llama 4 Scout and Qwen3 VL 235B A22B Instruct
SpecificationMeta Llama 4 ScoutQwen3 VL 235B A22B Instruct
MakerMeta LlamaQwen
WeightsOpen weightsOpen weights
ReleasedApril 5, 2025September 23, 2025
Context window1.31M262K
Input $/Mtok$0.175$0.21
Output $/Mtok$0.50$1.90
Coding indexNot benchmarkedNot benchmarked
Independent measurements behind it00
Cost per solved task$1.276Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is Llama 4 Scout better than Qwen3 VL 235B A22B Instruct for coding?

Neither model has a coding score yet, so the two cannot be ranked on coding. Pick Llama 4 Scout for output tokens 3.8x cheaper ($0.50 vs $1.90 per million) and a larger context window (1.31M against 262K tokens).

Which is cheaper, Llama 4 Scout or Qwen3 VL 235B A22B Instruct?

Llama 4 Scout costs $0.175 per million input tokens and $0.50 per million output tokens; Qwen3 VL 235B A22B Instruct costs $0.21 and $1.90. Cost per solved task is known for Llama 4 Scout only.

Which has the bigger context window, Llama 4 Scout or Qwen3 VL 235B A22B Instruct?

Llama 4 Scout, with 1.31M tokens against 262K.

How many benchmarks is the Llama 4 Scout vs Qwen3 VL 235B A22B Instruct comparison based on?

47 benchmarks have a result for both models as of October 22, 2026. Llama 4 Scout's coding index rests on 0 independent benchmarks and Qwen3 VL 235B A22B Instruct's on 0. Where only one model has a result, the page says so instead of naming a winner.

Compare next