LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

Claude Haiku 5.5 vs Qwen3.7 Max

Updated October 21, 2026. 33 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Claude Haiku 5.5 Replace Qwen3.7 Max
Claude Haiku 5.5Anthropic
Coding
75.9
Overall
43.4
Output $/M
$0.50
29
Qwen3.7 MaxQwen
Coding
45.1
Overall
29.5
Output $/M
$4.425
4
vs boards won of 33 shared

Decision reading

Pick Claude Haiku 5.5 for coding: it scores 75.9 on the Vibe Coding Index against 45.1 for Qwen3.7 Max.

Pick Qwen3.7 Max when what matters is GPQA Diamond (Epoch AI run) or Mystery game puzzles, where it leads.

Model AClaude Haiku 5.5Anthropic75.9/1002 independent benchmarksNo cost per solved taskModel BQwen3.7 MaxQwen45.1/1001 independent benchmark$4.69 per solved task
Shared results
33
Claude Haiku 5.5 only
29
Qwen3.7 Max only
26
Like-for-like categories
8 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
Claude Haiku 5.5 75.9 from 2 benchmarks, Qwen3.7 Max 45.1 from 1 benchmark.
Widest category
Knowledge, 22.4 points apart on average over 2 shared benchmarks.
Widest single gap
ProofBench: Claude Haiku 5.5 89.0% against Qwen3.7 Max 26.0%, 63.0 points apart on a 0 to 100 scale.
Output price
Claude Haiku 5.5 $0.50 against Qwen3.7 Max $4.425 per million output tokens, 8.8x apart.
Context window
Both take 1M tokens.

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
CategoryClaude Haiku 5.5BarsQwen3.7 MaxTies
Coding60
Intelligence20
Agentic work40
Reasoning32
Math40
Knowledge11
Web design51
General40
All shared boards294
CodingLike-for-likeFour of these boards feed the Vibe Coding Index60
Coding benchmarks for Claude Haiku 5.5 and Qwen3.7 Max
BenchmarkClaude Haiku 5.5Qwen3.7 MaxLeader
LiveBench coding91.4%Not measuredOnly Claude Haiku 5.5 measured
LMArena WebDev15861515Claude Haiku 5.5
AA Coding IndexNot measured66.0Only Qwen3.7 Max measured
LiveCodeBench (Vals AI)Not measured87.1%Only Qwen3.7 Max measured
SWE-bench Verified (Vals AI)Not measured68.8%Only Qwen3.7 Max measured
Terminal-Bench 4 (Vals AI)36.4%Not measuredOnly Claude Haiku 5.5 measured
Vibe Code Bench (Vals AI)90.4%47.7%Claude Haiku 5.5
Code migration (Vals AI)44.6%13.1%Claude Haiku 5.5
ProgramBench (Vals AI)2.5%Not measuredOnly Claude Haiku 5.5 measured
IOI (Vals AI)47.3%Not measuredOnly Claude Haiku 5.5 measured
Terminal-Bench 2.1 (Vals AI)Not measured61.0%Only Qwen3.7 Max measured
SWE-bench Multimodal (Vals AI)44.0%Not measuredOnly Claude Haiku 5.5 measured
Vibe Code Bench 1-100 (Vals AI)20.4%Not measuredOnly Claude Haiku 5.5 measured
CyberBench (Vals AI)75.5%Not measuredOnly Claude Haiku 5.5 measured
SciCode55.0%49.5%Claude Haiku 5.5
Terminal-Bench Hard (Artificial Analysis run)Not measured50.8%Only Qwen3.7 Max measured
FrontierCode46.4%Not measuredOnly Claude Haiku 5.5 measured
FrontierSWE43.8%Not measuredOnly Claude Haiku 5.5 measured
ALE-BenchNot measured1189.4Only Qwen3.7 Max measured
LMArena WebDev: React16051514Claude Haiku 5.5
LMArena WebDev: HTML15691512Claude Haiku 5.5
Neither model has a result on the other 17 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for Claude Haiku 5.5 and Qwen3.7 Max
BenchmarkClaude Haiku 5.5Qwen3.7 MaxLeader
AA Intelligence Index43.429.5Claude Haiku 5.5
Epoch Capabilities Index159.3154.0Claude Haiku 5.5
Agentic workLike-for-likePublic boards, shown as published, kept out of the index40
Agentic work benchmarks for Claude Haiku 5.5 and Qwen3.7 Max
BenchmarkClaude Haiku 5.5Qwen3.7 MaxLeader
AA Agentic IndexNot measured22.5Only Qwen3.7 Max measured
Design Arena: agents building web appsNot measured1208Only Qwen3.7 Max measured
Design Arena: agents building full-stack appsNot measured1191Only Qwen3.7 Max measured
Design Arena: agents building mobile appsNot measured1153Only Qwen3.7 Max measured
Design Arena: agents building Android appsNot measured1165Only Qwen3.7 Max measured
Design Arena: agents building Godot gamesNot measured1187Only Qwen3.7 Max measured
Design Arena: agents building gamesNot measured1161Only Qwen3.7 Max measured
Design Arena: HTML slidesNot measured1158Only Qwen3.7 Max measured
Design Arena: PowerPoint slidesNot measured1196Only Qwen3.7 Max measured
LiveBench research81.9%Not measuredOnly Claude Haiku 5.5 measured
LiveBench browser80.0%Not measuredOnly Claude Haiku 5.5 measured
LiveBench long agentic57.6%Not measuredOnly Claude Haiku 5.5 measured
SREBench (Vals AI)5.7%Not measuredOnly Claude Haiku 5.5 measured
Finance Agent (Vals AI)54.1%47.8%Claude Haiku 5.5
Tax Agent Bench (Vals AI)31.9%15.8%Claude Haiku 5.5
Legal Research Bench (Vals AI)43.3%25.5%Claude Haiku 5.5
Harvey's Legal Agent Benchmark (Vals AI)4.2%Not measuredOnly Claude Haiku 5.5 measured
Excel Modeling Benchmark (Vals AI)59.0%57.0%Claude Haiku 5.5
BioMysteryBench (Vals AI)70.4%Not measuredOnly Claude Haiku 5.5 measured
Terminal-Bench-Science (Vals AI)10.0%Not measuredOnly Claude Haiku 5.5 measured
Vals Science Index38.5%Not measuredOnly Claude Haiku 5.5 measured
τ²-Bench TelecomNot measured94.7%Only Qwen3.7 Max measured
APEX-Agents56.1%Not measuredOnly Claude Haiku 5.5 measured
Vending-Bench 2$3,602Not measuredOnly Claude Haiku 5.5 measured
Neither model has a result on the other 15 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index32
Reasoning benchmarks for Claude Haiku 5.5 and Qwen3.7 Max
BenchmarkClaude Haiku 5.5Qwen3.7 MaxLeader
GPQA DiamondNot measured90.2%Only Qwen3.7 Max measured
LiveBench reasoning82.4%Not measuredOnly Claude Haiku 5.5 measured
MysteryMechanism (Vals AI)35.1%Not measuredOnly Claude Haiku 5.5 measured
GPQA Diamond (Artificial Analysis run)Not measured92.3%Only Qwen3.7 Max measured
Humanity's Last Exam (Artificial Analysis run)44.4%40.5%Claude Haiku 5.5
GPQA Diamond (Epoch AI run)89.6%90.9%Qwen3.7 Max
ARC-AGI89.0%Not measuredOnly Claude Haiku 5.5 measured
ARC-AGI-261.3%Not measuredOnly Claude Haiku 5.5 measured
SimpleBenchNot measured70.4%Only Qwen3.7 Max measured
Chess puzzles55.0%26.0%Claude Haiku 5.5
Mystery game puzzles30.0%32.0%Qwen3.7 Max
EBR-bench43.8%9.5%Claude Haiku 5.5
Neither model has a result on the other 9 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index40
Math benchmarks for Claude Haiku 5.5 and Qwen3.7 Max
BenchmarkClaude Haiku 5.5Qwen3.7 MaxLeader
OTIS Mock AIME98.9%95.6%Claude Haiku 5.5
FrontierMath, tiers 1 to 375.1%64.6%Claude Haiku 5.5
FrontierMath, tier 446.3%34.1%Claude Haiku 5.5
ProofBench89.0%26.0%Claude Haiku 5.5
Neither model has a result on the other 9 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index11
Knowledge benchmarks for Claude Haiku 5.5 and Qwen3.7 Max
BenchmarkClaude Haiku 5.5Qwen3.7 MaxLeader
MMLU ProNot measured89.3%Only Qwen3.7 Max measured
LegalBench (Vals AI)Not measured84.9%Only Qwen3.7 Max measured
CorpFin (Vals AI)Not measured63.7%Only Qwen3.7 Max measured
TaxEval (Vals AI)Not measured75.3%Only Qwen3.7 Max measured
MedCode (Vals AI)51.5%38.8%Claude Haiku 5.5
SimpleQA Verified23.8%55.8%Qwen3.7 Max
Neither model has a result on the other 11 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index51
Web design benchmarks for Claude Haiku 5.5 and Qwen3.7 Max
BenchmarkClaude Haiku 5.5Qwen3.7 MaxLeader
Design Arena: websites12631274Qwen3.7 Max
Design Arena: UI components12951269Claude Haiku 5.5
Design Arena: game dev13151265Claude Haiku 5.5
Design Arena: data viz12921275Claude Haiku 5.5
Design Arena: 3D13431266Claude Haiku 5.5
Design Arena: all code categories12791275Claude Haiku 5.5
Design Arena: SVGNot measured1216Only Qwen3.7 Max measured
Design Arena: ASCII artNot measured1211Only Qwen3.7 Max measured
LiveBench design83.1%Not measuredOnly Claude Haiku 5.5 measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index40
General benchmarks for Claude Haiku 5.5 and Qwen3.7 Max
BenchmarkClaude Haiku 5.5Qwen3.7 MaxLeader
Vals Index58.7%Not measuredOnly Claude Haiku 5.5 measured
LiveBench data analysis53.1%Not measuredOnly Claude Haiku 5.5 measured
Public Benefits Bench (Vals AI)63.2%Not measuredOnly Claude Haiku 5.5 measured
SAGE (Vals AI)45.8%Not measuredOnly Claude Haiku 5.5 measured
MedScribe (Vals AI)89.5%79.4%Claude Haiku 5.5
IFBenchNot measured80.5%Only Qwen3.7 Max measured
Long context reasoning (AA-LCR)82.7%79.0%Claude Haiku 5.5
DTBench95.0%92.3%Claude Haiku 5.5
LMCA51.1%44.0%Claude Haiku 5.5
Furniture assembly47.5%Not measuredOnly Claude Haiku 5.5 measured
Neither model has a result on the other 36 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Claude Haiku 5.5 and Qwen3.7 Max
BenchmarkClaude Haiku 5.5Qwen3.7 MaxLeader
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

Claude Haiku 5.5$0.0003Fits in one request
Qwen3.7 Max$0.0037Fits in one request

Claude Haiku 5.5 costs less

Repository review

50K input + 3K output tokens

Claude Haiku 5.5$0.0065Fits in one request
Qwen3.7 Max$0.087Fits in one request

Claude Haiku 5.5 costs less

Agent loop

200K input + 10K output tokens

Claude Haiku 5.5$0.025Fits in one request
Qwen3.7 Max$0.339Fits in one request

Claude Haiku 5.5 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 Claude Haiku 5.5 and Qwen3.7 Max
SpecificationAnthropic Claude Haiku 5.5Qwen3.7 Max
MakerAnthropicQwen
WeightsNo public weights listedNo public weights listed
ReleasedOctober 7, 2026May 21, 2026
Context window1M1M
Input $/Mtok$0.10$1.475
Output $/Mtok$0.50$4.425
Coding index75.945.1
Independent measurements behind it21
Cost per solved taskNot computable$4.69
SourceOpenRouter listingOpenRouter listing

Questions

Is Claude Haiku 5.5 better than Qwen3.7 Max for coding?

Pick Claude Haiku 5.5 for coding: it scores 75.9 on the Vibe Coding Index against 45.1 for Qwen3.7 Max. Pick Qwen3.7 Max when what matters is GPQA Diamond (Epoch AI run) or Mystery game puzzles, where it leads.

Which is cheaper, Claude Haiku 5.5 or Qwen3.7 Max?

Claude Haiku 5.5 costs $0.10 per million input tokens and $0.50 per million output tokens; Qwen3.7 Max costs $1.475 and $4.425. Cost per solved task is known for Qwen3.7 Max only.

Which has the bigger context window, Claude Haiku 5.5 or Qwen3.7 Max?

Neither: both take 1M tokens.

How many benchmarks is the Claude Haiku 5.5 vs Qwen3.7 Max comparison based on?

33 benchmarks have a result for both models as of October 21, 2026. Claude Haiku 5.5's coding index rests on 2 independent benchmarks and Qwen3.7 Max's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next