Skip to main content
BenchLM
Data

Claude Haiku 5.5 vs Qwen3.8 Max

Updated November 21, 2026. Rank says Qwen3.8 Max is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.

Share or export
Share on XLinkedInAPI/MCP

Decision reading

Qwen3.8 Max has the higher public score estimate, 65.3 versus 63.5, but the 90% score intervals overlap. Treat that as a lead, not a settled winner. 10 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Model A
Anthropic logo

Anthropic

63.5/100

Supported · Public rank #38

90% interval 59.5–67.6

Model B
Alibaba logo

Alibaba

65.3/100

Supported · Public rank #36

90% interval 61.3–69.4

API price per 1M tokens

Claude Haiku 5.5: $0.10 input / $0.50 output

Qwen3.8 Max: No comparable API price

Context window

Claude Haiku 5.5: 1M

Qwen3.8 Max: 1M

Released

Claude Haiku 5.5: October 7, 2026

Qwen3.8 Max: August 3, 2026

Shared results
10
Claude Haiku 5.5 only
17
Qwen3.8 Max only
60
Like-for-like categories
7 / 8
Supported: Claude Haiku 5.5 and Qwen3.8 MaxHow the comparison works

Which one for your work

Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.

  • Agentic work

    Tool use, computer use, and multi-step task completion

    Claude Haiku 5.5

    Claude Haiku 5.5 has the higher public agentic point estimate, 68.3 to 64.8, with Supported evidence for both models. The conditional ranges do not establish rank confidence.

    Confidence: limited
Show secondary and unsupported calls
  • Coding work

    Code generation, repair, and software-engineering tasks

    No clear pick

    The like-for-like coding result is a practical tie on the public lane (within 0.5 points).

    Confidence: limited
  • Long documents

    Prompts that approach the documented context limit

    No clear pick

    The documented context windows are equal.

    Confidence: documented
  • Chat turn cost

    1K fresh input + 500 output tokens

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates
  • Cache-heavy agent loop cost

    200K cached + 20K fresh input + 10K output tokens

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates
  • Repository review cost

    50K fresh input + 3K output tokens

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates

Which one for a specific job

Choose a job from the LLM Selector's task catalog to see the category row it rests on, under the same basis rules as the table below. A directional row stays directional; choosing a job never creates a winner.

The same task catalog as the LLM Selector. Each job names the evidence surface it rests on; nothing here adds a new score.

67.9Claude Haiku 5.567.5Qwen3.8 Max

Like-for-like · BenchAlign v6.0

Claude Haiku 5.5 has the higher coding point estimate. Conditional score ranges do not establish rank confidence.

Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.

Same basis rules as the category table below

Or describe your own job

The LLM Selector reads what you need a model for and returns a shortlist, ranked in the same order the job pages publish.

Describe the task and constraints. This selector recommends models; it cannot carry out the task or access your files and accounts.

We keep what you type here, drafts included, for 180 days, and after that only the job and requirements read from it. Nothing is tied to your IP address or account. Leave out names and other personal details. Privacy

What is actually comparable

Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.

Shape of the matched evidence

Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.

Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.

Bars run 0–100 on each benchmark’s normalized display scale

Category results, on a stated basis

Each row shows the public-lane category score for both models: the BenchAlign v6.0 lane in every category, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence. Directional and not-comparable rows remain visible, but they never receive a winner.

Agentic

Like-for-like
Claude Haiku 5.5
68.3
Supported · #25/182
Qwen3.8 Max
64.8
Supported · #38/182
Basis
BenchAlign v6.0 lane · 9 vs 20 public rows
Reading
Claude Haiku 5.5 leads · intervals overlap

Coding

Like-for-like
Claude Haiku 5.5
67.9
Supported · #28/180
Qwen3.8 Max
67.5
Supported · #30/180
Basis
BenchAlign v6.0 lane · 7 vs 15 public rows
Reading
Practical tie

Reasoning

Like-for-like
Claude Haiku 5.5
62.6
Supported · #38/161
Qwen3.8 Max
64.7
Supported · #31/161
Basis
BenchAlign v6.0 lane · 4 vs 2 public rows
Reading
Qwen3.8 Max leads · intervals overlap

Multimodal

Like-for-like
Claude Haiku 5.5
63.9
Supported · #39/119
Qwen3.8 Max
65.6
Supported · #29/119
Basis
BenchAlign v6.0 lane · 4 vs 24 public rows
Reading
Qwen3.8 Max leads · intervals overlap

Knowledge

Like-for-like
Claude Haiku 5.5
50.2
Supported · #84/158
Qwen3.8 Max
60.5
Supported · #41/158
Basis
BenchAlign v6.0 lane · 1 vs 6 public rows
Reading
Qwen3.8 Max leads · intervals overlap

Instruction following

Like-for-like
Claude Haiku 5.5
58.3
Supported · #37/143
Qwen3.8 Max
60.9
Supported · #33/143
Basis
BenchAlign v6.0 lane · 0 vs 1 public rows
Reading
Qwen3.8 Max leads · intervals overlap

Math

Like-for-like
Claude Haiku 5.5
62.1
Supported · #29/125
Qwen3.8 Max
62.6
Supported · #27/125
Basis
BenchAlign v6.0 lane · 2 vs 2 public rows
Reading
Qwen3.8 Max leads · intervals overlap

Multilingual

Not comparable
Claude Haiku 5.5
56.1
Estimated · #36/88
Qwen3.8 Max
Not ranked
Basis
BenchAlign v6.0 lane · 0 vs 0 public rows
Reading
Not comparable

Ranks count the models scored in each category’s BenchAlign v6.0 lane, so denominators differ by category. Like-for-like rows within 0.5 points read as a practical tie.

Supported evidence per lane · bars run 0–100Methodology

What each workload costs

Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.

Chat turn

1K fresh input + 500 output tokens

Claude Haiku 5.5
$0.00035
Fits in one request
Qwen3.8 Max
API rate not published
Fits in one request

Qwen3.8 Max has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

Claude Haiku 5.5
$0.0065
Fits in one request
Qwen3.8 Max
API rate not published
Fits in one request

Qwen3.8 Max has no comparable published API token rate.

Cache-heavy agent loop

200K cached + 20K fresh input + 10K output tokens

Claude Haiku 5.5
$0.009
Fits in one request
Qwen3.8 Max
API rate not published
Fits in one request
Cached-input rate unavailable

Qwen3.8 Max has no comparable published API token rate.

Cached input falls back to the list input rate only where a cached rate is unpublished

Price against score, among all priced models

Overall score against output price per 1M tokens, one model per family. Pricing as of November 21, 2026.

Claude Haiku 5.5

20406080100$0.10$0.50$1$5$10$50$100Output price, $ per 1M tokens (log)Overall score
  • Claude Haiku 5.5 is on the efficiency frontier: no cheaper model scores at least as high.
  • Qwen3.8 Max has no public API price, so it is not plotted.

Gray dots are the other priced models. The dark step line is the efficiency frontier: for a model under it, another model costs no more and scores at least as high.

See Claude Haiku 5.5 on the full price vs performance chart

Specification differences

Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.

Reasoning profile

Claude Haiku 5.5

Reasoning

Qwen3.8 Max

Reasoning

Weight access

Claude Haiku 5.5

Proprietary

Qwen3.8 Max

Open Weight

License

Claude Haiku 5.5

Proprietary

Qwen3.8 Max

Open Weight

Release date

Claude Haiku 5.5

2026-10-07

Qwen3.8 Max

2026-08-03

If you already use one of these models

Deployment change
The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
Quality signal
Qwen3.8 Max has the higher public score estimate, 65.3 versus 63.5, but the 90% score intervals overlap.
Workload cost
A complete comparable API-rate estimate is not available for both models.
Context tradeoff
Both models list 1M.
Run the same representative tasks against both endpoints before changing production traffic.

Questions

Which is better, Claude Haiku 5.5 or Qwen3.8 Max?

Qwen3.8 Max has the higher public score estimate, 65.3 versus 63.5, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.

Which is better for coding, Claude Haiku 5.5 or Qwen3.8 Max?

The like-for-like coding row is a practical tie on the public lane, 67.9 against 67.5, inside the 0.5-point band BenchLM treats as level.

Which is better for agentic tasks, Claude Haiku 5.5 or Qwen3.8 Max?

Claude Haiku 5.5 has the higher public agentic tasks point estimate, 68.3 to 64.8, with Supported evidence for both models. The conditional ranges do not establish rank confidence.

Which costs less, Claude Haiku 5.5 or Qwen3.8 Max?

Both models do not have comparable published API token rates, so this page does not name a universal price winner.

Which has the larger context window, Claude Haiku 5.5 or Qwen3.8 Max?

Both models list the same context window, 1M.

Self-host vs API cost

Estimates at 50,000 req/day · 1000 tokens/req average.

Claude Haiku 5.5
API / mo$450
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
Qwen3.8 Max
API / mo (DeepInfra)$4,951
Self-host / mo$57,606
Break-evenNot reached at 70% load
Model the full break-even

Benchmark evidence

The full public result ledger is available for audit without forcing a wide desktop table onto a phone.

Browse raw public benchmark evidence87 rows

Agentic

  • Claude Haiku 5.558.62%
    Qwen3.8 Max55.66%

    Claude Haiku 5.5 leads this result

  • Terminal-Bench 4.1

    Claude Haiku 5.540.00%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • Agents' Last Exam (Linux)

    Claude Haiku 5.557.0%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • Terminal-Bench 4.0

    Claude Haiku 5.539.20%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • HLE w/ tools

    Claude Haiku 5.557.4%
    Source
    Qwen3.8 Max56.2%
    Source

    Claude Haiku 5.5 leads this result

  • MazeBench

    Claude Haiku 5.51%
    Source
    Qwen3.8 Max0%
    Source

    Claude Haiku 5.5 leads this result

  • Hermes Index

    Shared source
    Claude Haiku 5.537.0%
    Qwen3.8 Max36.2%

    Claude Haiku 5.5 leads this result

  • Hermes Bench

    Shared source
    Claude Haiku 5.571.0%
    Qwen3.8 Max70.5%

    Claude Haiku 5.5 leads this result

  • OSWorld 2.1 Offline

    Claude Haiku 5.572.4%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • Terminal-Bench 2.1

    Claude Haiku 5.5—
    Qwen3.8 Max86.6%
    Source

    Not directly comparable

  • CoWorkBench

    Claude Haiku 5.5—
    Qwen3.8 Max74.8%
    Source

    Not directly comparable

  • JobBench

    Claude Haiku 5.5—
    Qwen3.8 Max53.4%
    Source

    Not directly comparable

  • skillsBench

    Claude Haiku 5.5—
    Qwen3.8 Max70.2%
    Source

    Not directly comparable

  • Agents' Last Exam

    Claude Haiku 5.5—
    Qwen3.8 Max52.4%
    Source

    Not directly comparable

  • AutomationBench

    Claude Haiku 5.5—
    Qwen3.8 Max27.3%
    Source

    Not directly comparable

  • Toolathlon-Verified

    Claude Haiku 5.5—
    Qwen3.8 Max72.5%
    Source

    Not directly comparable

  • WideResearch

    Claude Haiku 5.5—
    Qwen3.8 Max81.9%
    Source

    Not directly comparable

  • OSWorld-Verified

    Claude Haiku 5.5—
    Qwen3.8 Max86.1%
    Source

    Not directly comparable

  • OSWorld 2.0

    Claude Haiku 5.5—
    Qwen3.8 Max19.4%
    Source

    Not directly comparable

  • WebArena-Verified

    Claude Haiku 5.5—
    Qwen3.8 Max66.8%
    Source

    Not directly comparable

  • AndroidWorld

    Claude Haiku 5.5—
    Qwen3.8 Max85.3%
    Source

    Not directly comparable

  • MobileWorld

    Claude Haiku 5.5—
    Qwen3.8 Max77.8%
    Source

    Not directly comparable

  • Terminal-Bench 2.1 (Vals)

    Claude Haiku 5.5—
    Qwen3.8 Max67.4%
    Source

    Not directly comparable

  • CWE-bench

    Claude Haiku 5.5—
    Qwen3.8 Max45.4%
    Source

    Not directly comparable

Coding

  • FrontierCode 1.1 Main

    Claude Haiku 5.546.4%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • Bug Hunt Bench

    Shared source
    Claude Haiku 5.521.5 fixes
    Qwen3.8 Max25.7 fixes

    Qwen3.8 Max leads this result

  • FrontierSWE v2

    Shared source
    Claude Haiku 5.543.8%
    Qwen3.8 Max15.8%

    Claude Haiku 5.5 leads this result

  • FrontierCode 2.0

    Claude Haiku 5.528.9%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • CursorBench 4.0

    Claude Haiku 5.548.4%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • VulcanBench Frontier v4

    Claude Haiku 5.592.2%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • CursorBench 4.1

    Claude Haiku 5.548.4%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • Terminal-Bench 2.1

    Claude Haiku 5.5—
    Qwen3.8 Max86.6%
    Source

    Not directly comparable

  • SWE-bench Pro

    Claude Haiku 5.5—
    Qwen3.8 Max67.7%
    Source

    Not directly comparable

  • DeepSWE

    Claude Haiku 5.5—
    Qwen3.8 Max56.6%
    Source

    Not directly comparable

  • NL2Repo

    Claude Haiku 5.5—
    Qwen3.8 Max55.9%
    Source

    Not directly comparable

  • FrontierSWE

    Claude Haiku 5.5—
    Qwen3.8 Max73.5%
    Source

    Not directly comparable

  • MLS-Bench Lite

    Claude Haiku 5.5—
    Qwen3.8 Max41.0%
    Source

    Not directly comparable

  • PaperBench

    Claude Haiku 5.5—
    Qwen3.8 Max93.0%
    Source

    Not directly comparable

  • VulcanBench v3

    Claude Haiku 5.5—
    Qwen3.8 Max81.2%
    Source

    Not directly comparable

  • OpenHarmony Bench v1.0

    Claude Haiku 5.5—
    Qwen3.8 Max60.8%
    Source

    Not directly comparable

  • LiveCodeBench (Vals)

    Claude Haiku 5.5—
    Qwen3.8 Max87.9%
    Source

    Not directly comparable

  • SWE-bench (Vals)

    Claude Haiku 5.5—
    Qwen3.8 Max85.6%
    Source

    Not directly comparable

  • ScarfBench

    Claude Haiku 5.5—
    Qwen3.8 Max63.7%
    Source

    Not directly comparable

  • OpenHarmony Bench v1.2

    Claude Haiku 5.5—
    Qwen3.8 Max67.77%
    Source

    Not directly comparable

Reasoning

  • ARC-AGI-1

    Claude Haiku 5.589.00%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • ARC-AGI-2

    Claude Haiku 5.561.3%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • ARC-AGI-3

    Claude Haiku 5.50.5%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • ARC-AGI-3 Provider Adapter

    Claude Haiku 5.52.82%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • MRCRv2

    Claude Haiku 5.5—
    Qwen3.8 Max92.9%
    Source

    Not directly comparable

  • LongBench v2

    Claude Haiku 5.5—
    Qwen3.8 Max66.3%
    Source

    Not directly comparable

Multimodal

  • ParseBench-2

    Claude Haiku 5.571.24%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • PerceptionBench (tools)

    Claude Haiku 5.578.6%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • ParseBench-1

    Claude Haiku 5.556.06%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • Chartography (no tools)

    Claude Haiku 5.546.4%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • MMMU-Pro

    Claude Haiku 5.5—
    Qwen3.8 Max82.3%
    Source

    Not directly comparable

  • MathVision

    Claude Haiku 5.5—
    Qwen3.8 Max95.2%
    Source

    Not directly comparable

  • MathVision w/ Python

    Claude Haiku 5.5—
    Qwen3.8 Max97.7%
    Source

    Not directly comparable

  • BabyVision

    Claude Haiku 5.5—
    Qwen3.8 Max82.0%
    Source

    Not directly comparable

  • BabyVision w/ Python

    Claude Haiku 5.5—
    Qwen3.8 Max91.3%
    Source

    Not directly comparable

  • ZeroBench

    Claude Haiku 5.5—
    Qwen3.8 Max24.0%
    Source

    Not directly comparable

  • ZeroBench w/ Python

    Claude Haiku 5.5—
    Qwen3.8 Max49.0%
    Source

    Not directly comparable

  • MedXpertQA (MM)

    Claude Haiku 5.5—
    Qwen3.8 Max80.4%
    Source

    Not directly comparable

  • ScreenSpot Pro

    Claude Haiku 5.5—
    Qwen3.8 Max84.5%
    Source

    Not directly comparable

  • Vision2Web

    Claude Haiku 5.5—
    Qwen3.8 Max69.0%
    Source

    Not directly comparable

  • CharXiv w/o tools

    Claude Haiku 5.5—
    Qwen3.8 Max88.4%
    Source

    Not directly comparable

  • CharXiv

    Claude Haiku 5.5—
    Qwen3.8 Max93.5%
    Source

    Not directly comparable

  • OmniDocBench 1.5

    Claude Haiku 5.5—
    Qwen3.8 Max92.1%
    Source

    Not directly comparable

  • OCRBench V2

    Claude Haiku 5.5—
    Qwen3.8 Max74.2%
    Source

    Not directly comparable

  • CC-OCR

    Claude Haiku 5.5—
    Qwen3.8 Max79.6%
    Source

    Not directly comparable

  • RealWorldQA

    Claude Haiku 5.5—
    Qwen3.8 Max88.0%
    Source

    Not directly comparable

  • ERQA

    Claude Haiku 5.5—
    Qwen3.8 Max77.8%
    Source

    Not directly comparable

  • SimpleVQA

    Claude Haiku 5.5—
    Qwen3.8 Max75.0%
    Source

    Not directly comparable

  • PerceptionBench

    Claude Haiku 5.5—
    Qwen3.8 Max63.5%
    Source

    Not directly comparable

  • Video-MME (with subtitle)

    Claude Haiku 5.5—
    Qwen3.8 Max90.4%
    Source

    Not directly comparable

  • VideoMMMU

    Claude Haiku 5.5—
    Qwen3.8 Max88.7%
    Source

    Not directly comparable

  • MMVU

    Claude Haiku 5.5—
    Qwen3.8 Max82.4%
    Source

    Not directly comparable

  • MLVU (M-Avg)

    Claude Haiku 5.5—
    Qwen3.8 Max90.8%
    Source

    Not directly comparable

  • LVBench

    Claude Haiku 5.5—
    Qwen3.8 Max81.8%
    Source

    Not directly comparable

Knowledge

  • HLE w/o tools

    Claude Haiku 5.545.9%
    Source
    Qwen3.8 Max43.6%
    Source

    Claude Haiku 5.5 leads this result

  • GPQA

    Claude Haiku 5.5—
    Qwen3.8 Max92.6%
    Source

    Not directly comparable

  • GPQA-D

    Claude Haiku 5.5—
    Qwen3.8 Max92.6%
    Source

    Not directly comparable

  • HLE

    Claude Haiku 5.5—
    Qwen3.8 Max43.6%
    Source

    Not directly comparable

  • GPQA Diamond (Vals)

    Claude Haiku 5.5—
    Qwen3.8 Max93.7%
    Source

    Not directly comparable

  • MMLU-Pro (Vals)

    Claude Haiku 5.5—
    Qwen3.8 Max88.6%
    Source

    Not directly comparable

Instruction following

  • IFBench

    Claude Haiku 5.5—
    Qwen3.8 Max82.8%
    Source

    Not directly comparable

Math

  • FrontierMath v2 (Tiers 1-3)

    Shared source
    Claude Haiku 5.575.088%
    Qwen3.8 Max74.737%

    Claude Haiku 5.5 leads this result

  • FrontierMath v2 (Tier 4)

    Shared source
    Claude Haiku 5.546.341%
    Qwen3.8 Max46.341%

    Tie

87 public results · 10 shared

Watch Claude Haiku 5.5 vs Qwen3.8 Max

One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.

Read a sample issue

Join 7,000+ subscribers.

Last updated November 21, 2026