Enter opens the highlighted result. Esc closes.

Grok 4.3 vs Qwen3.5-27B

Updated October 21, 2026. 52 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Grok 4.3 Replace Qwen3.5-27B
Grok 4.3xAI
Coding
22.5
Overall
24.9
Output $/M
$2.50
21
Qwen3.5-27BQwen
Coding
22.6
Overall
22.9
Output $/M
$2.16
31
vs boards won of 52 shared

Decision reading

Grok 4.3 and Qwen3.5-27B are level for coding, 22.5 against 22.6 on the Vibe Coding Index, so price and context decide.

Pick Grok 4.3 when the context window is larger, 1M against 262K tokens, or when what matters is Terminal-Bench Hard (Artificial Analysis run) or WeirdML, where it leads. Pick Qwen3.5-27B when output tokens cost 1.2x less, or when what matters is LMArena WebDev or LMArena Text: coding, where it leads.

Model AGrok 4.3xAI22.5/1001 independent benchmark$0.31 per solved taskModel BQwen3.5-27BQwen22.6/1001 independent benchmarkNo cost per solved task
Shared results
52
Grok 4.3 only
52
Qwen3.5-27B only
1
Like-for-like categories
7 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
Grok 4.3 22.5 from 1 benchmark, Qwen3.5-27B 22.6 from 1 benchmark.
Widest category
Math, 6.3 points apart on average over 2 shared benchmarks.
Widest single gap
ALE-Bench: Grok 4.3 944.2 against Qwen3.5-27B 349.4, 19.8 points apart on a 0 to 100 scale.
Output price
Grok 4.3 $2.50 against Qwen3.5-27B $2.16 per million output tokens, 1.2x apart.
Context window
Grok 4.3 takes 1M tokens, Qwen3.5-27B 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
CategoryGrok 4.3BarsQwen3.5-27BTies
Coding44
Intelligence10
Agentic work11
Reasoning25
Math02
Knowledge04
General1315
All shared boards2131
CodingLike-for-likeFour of these boards feed the Vibe Coding Index44
Coding benchmarks for Grok 4.3 and Qwen3.5-27B
BenchmarkGrok 4.3Qwen3.5-27BLeader
LMArena WebDev13581358Qwen3.5-27B
AA Coding Index42.2Not measuredOnly Grok 4.3 measured
LiveCodeBench (Vals AI)84.5%Not measuredOnly Grok 4.3 measured
SWE-bench Verified (Vals AI)71.4%Not measuredOnly Grok 4.3 measured
Vibe Code Bench (Vals AI)19.4%Not measuredOnly Grok 4.3 measured
Code migration (Vals AI)6.8%Not measuredOnly Grok 4.3 measured
ProgramBench (Vals AI)0.0%Not measuredOnly Grok 4.3 measured
Terminal-Bench 2.1 (Vals AI)42.0%Not measuredOnly Grok 4.3 measured
SkillsBench (Vals AI)40.6%Not measuredOnly Grok 4.3 measured
SciCode48.3%Not measuredOnly Grok 4.3 measured
Terminal-Bench Hard (Artificial Analysis run)37.9%32.6%Grok 4.3
WeirdML49.9%39.5%Grok 4.3
ALE-Bench944.2349.4Grok 4.3
LMArena Text: coding14151427Qwen3.5-27B
LMArena Text: software and IT14161431Qwen3.5-27B
LMArena WebDev: React13531346Grok 4.3
LMArena WebDev: HTML13671386Qwen3.5-27B
LMArena WebDev: from an image1375Not measuredOnly Grok 4.3 measured
Neither model has a result on the other 20 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
Intelligence benchmarks for Grok 4.3 and Qwen3.5-27B
BenchmarkGrok 4.3Qwen3.5-27BLeader
AA Intelligence Index24.922.9Grok 4.3
Neither model has a result on the other 1 board here
Agentic workLike-for-likePublic boards, shown as published, kept out of the index11
Agentic work benchmarks for Grok 4.3 and Qwen3.5-27B
BenchmarkGrok 4.3Qwen3.5-27BLeader
AA Agentic Index15.5Not measuredOnly Grok 4.3 measured
Design Arena: agents building web apps1126Not measuredOnly Grok 4.3 measured
Design Arena: agents building full-stack apps1006Not measuredOnly Grok 4.3 measured
Design Arena: agents building mobile apps1059Not measuredOnly Grok 4.3 measured
Design Arena: agents building Android apps967Not measuredOnly Grok 4.3 measured
Design Arena: agents building Godot games914Not measuredOnly Grok 4.3 measured
Design Arena: agents building games1007Not measuredOnly Grok 4.3 measured
Design Arena: HTML slides1008Not measuredOnly Grok 4.3 measured
Design Arena: PowerPoint slides1012Not measuredOnly Grok 4.3 measured
Finance Agent (Vals AI)37.7%Not measuredOnly Grok 4.3 measured
Tax Agent Bench (Vals AI)6.2%Not measuredOnly Grok 4.3 measured
Legal Research Bench (Vals AI)15.4%Not measuredOnly Grok 4.3 measured
Excel Modeling Benchmark (Vals AI)18.4%Not measuredOnly Grok 4.3 measured
τ²-Bench Telecom97.7%93.9%Grok 4.3
Vending-Bench 2$35$202Qwen3.5-27B
LMArena Search1166Not measuredOnly Grok 4.3 measured
Neither model has a result on the other 23 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index25
Reasoning benchmarks for Grok 4.3 and Qwen3.5-27B
BenchmarkGrok 4.3Qwen3.5-27BLeader
GPQA Diamond91.4%Not measuredOnly Grok 4.3 measured
MMMU Pro (Vals AI)83.1%Not measuredOnly Grok 4.3 measured
GPQA Diamond (Artificial Analysis run)90.1%85.8%Grok 4.3
Humanity's Last Exam (Artificial Analysis run)37.2%23.9%Grok 4.3
GPQA Diamond (Epoch AI run)88.8%Not measuredOnly Grok 4.3 measured
Chess puzzles18.0%Not measuredOnly Grok 4.3 measured
LMArena Text: hard prompts13961415Qwen3.5-27B
LMArena Text: expert13841427Qwen3.5-27B
LMArena Text: hard prompts in English14021429Qwen3.5-27B
LMArena Vision: diagrams12411249Qwen3.5-27B
LMArena Vision: homework12411256Qwen3.5-27B
Neither model has a result on the other 10 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for Grok 4.3 and Qwen3.5-27B
BenchmarkGrok 4.3Qwen3.5-27BLeader
OTIS Mock AIME93.3%Not measuredOnly Grok 4.3 measured
FrontierMath, tiers 1 to 342.8%Not measuredOnly Grok 4.3 measured
FrontierMath, tier 414.6%Not measuredOnly Grok 4.3 measured
ProofBench11.0%Not measuredOnly Grok 4.3 measured
LMArena Text: math13871429Qwen3.5-27B
LMArena Text: mathematical work13961429Qwen3.5-27B
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index04
Knowledge benchmarks for Grok 4.3 and Qwen3.5-27B
BenchmarkGrok 4.3Qwen3.5-27BLeader
MMLU Pro85.8%Not measuredOnly Grok 4.3 measured
LegalBench (Vals AI)84.5%Not measuredOnly Grok 4.3 measured
CaseLaw (Vals AI)79.3%Not measuredOnly Grok 4.3 measured
CorpFin (Vals AI)68.5%Not measuredOnly Grok 4.3 measured
TaxEval (Vals AI)70.8%Not measuredOnly Grok 4.3 measured
MortgageTax (Vals AI)48.3%Not measuredOnly Grok 4.3 measured
MedCode (Vals AI)38.1%Not measuredOnly Grok 4.3 measured
SimpleQA Verified33.2%Not measuredOnly Grok 4.3 measured
LMArena Text: science13991437Qwen3.5-27B
LMArena Text: medicine and healthcare13841425Qwen3.5-27B
LMArena Text: legal and government13861405Qwen3.5-27B
LMArena Text: business and finance13811399Qwen3.5-27B
Neither model has a result on the other 5 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
Web design benchmarks for Grok 4.3 and Qwen3.5-27B
BenchmarkGrok 4.3Qwen3.5-27BLeader
Design Arena: websites1196Not measuredOnly Grok 4.3 measured
Design Arena: UI components1192Not measuredOnly Grok 4.3 measured
Design Arena: game dev1174Not measuredOnly Grok 4.3 measured
Design Arena: data viz1180Not measuredOnly Grok 4.3 measured
Design Arena: 3D1125Not measuredOnly Grok 4.3 measured
Design Arena: all code categories1187Not measuredOnly Grok 4.3 measured
Design Arena: SVG1081Not measuredOnly Grok 4.3 measured
Design Arena: ASCII art1125Not measuredOnly Grok 4.3 measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index1315
General benchmarks for Grok 4.3 and Qwen3.5-27B
BenchmarkGrok 4.3Qwen3.5-27BLeader
Public Benefits Bench (Vals AI)51.7%Not measuredOnly Grok 4.3 measured
SAGE (Vals AI)19.7%Not measuredOnly Grok 4.3 measured
MedScribe (Vals AI)74.4%Not measuredOnly Grok 4.3 measured
Vals Multimodal Index43.3%Not measuredOnly Grok 4.3 measured
IFBench83.3%75.6%Grok 4.3
Long context reasoning (AA-LCR)75.0%77.7%Qwen3.5-27B
DTBench90.7%82.4%Grok 4.3
LMCA38.3%34.0%Grok 4.3
LMArena Text13971409Qwen3.5-27B
LMArena Text: instruction following13661393Qwen3.5-27B
LMArena Text: creative writing13801362Grok 4.3
LMArena Vision12291241Qwen3.5-27B
LMArena Text: English14041425Qwen3.5-27B
LMArena Text: non-English13851390Qwen3.5-27B
LMArena Text: Chinese14221478Qwen3.5-27B
LMArena Text: German13951392Grok 4.3
LMArena Text: Japanese13791346Grok 4.3
LMArena Text: Korean13541359Qwen3.5-27B
LMArena Text: French14111410Grok 4.3
LMArena Text: Spanish13991407Qwen3.5-27B
LMArena Text: Russian13991390Grok 4.3
LMArena Text: Polish13981375Grok 4.3
LMArena Text: multi-turn14061410Qwen3.5-27B
LMArena Text: longer queries13931413Qwen3.5-27B
LMArena Text: writing and language13851380Grok 4.3
LMArena Text: entertainment and media13701363Grok 4.3
LMArena Vision: English12351246Qwen3.5-27B
LMArena Vision: Chinese12571301Qwen3.5-27B
LMArena Vision: reading text12351246Qwen3.5-27B
LMArena Vision: humor12221227Qwen3.5-27B
LMArena Vision: captioningNot measured1170Only Qwen3.5-27B measured
LMArena Vision: recognizing things12171210Grok 4.3
LMArena Vision: creative writing12371224Grok 4.3
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 Grok 4.3 and Qwen3.5-27B
BenchmarkGrok 4.3Qwen3.5-27BLeader
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

Grok 4.3$0.0025Fits in one request
Qwen3.5-27B$0.0014Fits in one request

Qwen3.5-27B costs less

Repository review

50K input + 3K output tokens

Grok 4.3$0.070Fits in one request
Qwen3.5-27B$0.020Fits in one request

Qwen3.5-27B costs less

Agent loop

200K input + 10K output tokens

Grok 4.3$0.275Fits in one request
Qwen3.5-27B$0.076Fits in one request

Qwen3.5-27B 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 Grok 4.3 and Qwen3.5-27B
SpecificationSpaceXAI Grok 4.3Qwen3.5-27B
MakerxAIQwen
WeightsNo public weights listedOpen weights
ReleasedApril 30, 2026February 25, 2026
Context window1M262K
Input $/Mtok$1.25$0.27
Output $/Mtok$2.50$2.16
Coding index22.522.6
Independent measurements behind it11
Cost per solved task$0.31Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is Grok 4.3 better than Qwen3.5-27B for coding?

Grok 4.3 and Qwen3.5-27B are level for coding, 22.5 against 22.6 on the Vibe Coding Index, so price and context decide. Pick Grok 4.3 when the context window is larger, 1M against 262K tokens, or when what matters is Terminal-Bench Hard (Artificial Analysis run) or WeirdML, where it leads. Pick Qwen3.5-27B when output tokens cost 1.2x less, or when what matters is LMArena WebDev or LMArena Text: coding, where it leads.

Which is cheaper, Grok 4.3 or Qwen3.5-27B?

Grok 4.3 costs $1.25 per million input tokens and $2.50 per million output tokens; Qwen3.5-27B costs $0.27 and $2.16. Cost per solved task is known for Grok 4.3 only.

Which has the bigger context window, Grok 4.3 or Qwen3.5-27B?

Grok 4.3, with 1M tokens against 262K.

How many benchmarks is the Grok 4.3 vs Qwen3.5-27B comparison based on?

52 benchmarks have a result for both models as of October 21, 2026. Grok 4.3's coding index rests on 1 independent benchmark and Qwen3.5-27B's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next