Skip to main content
BenchLM

Qwen3.8-27B vs Qwen3.8 Max

Updated September 28, 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 XLinkedInSocial cardCSVJSON

Decision reading

Qwen3.8 Max has the higher public score, 71.69 versus 55.26, and the 90% score intervals do not overlap. 32 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Model A
Alibaba logo
Qwen3.8-27B

Alibaba

55.26/100

Estimated · Public rank #54

90% interval 47.7–62.8

Model B
Alibaba logo
Qwen3.8 Max

Alibaba

71.69/100

Supported · Public rank #13

90% interval 68.5–74.8

Shared results
32
Qwen3.8-27B only
1
Qwen3.8 Max only
28
Like-for-like categories
4 / 8
Estimated: Qwen3.8-27B · Supported: 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.

  • Coding work

    Code generation, repair, and software-engineering tasks

    Qwen3.8 Max

    Qwen3.8 Max leads on the public coding lane, 55.6 to 48.7, with Supported evidence for both models, although the 90% intervals overlap.

    Confidence: limited

  • Agentic work

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

    Qwen3.8 Max

    Qwen3.8 Max leads on the public agentic lane, 64.8 to 61.2, with Supported evidence for both models, although the 90% intervals overlap.

    Confidence: limited

  • Long documents

    Prompts that approach the documented context limit

    Qwen3.8 Max

    Qwen3.8 Max has the larger documented context window.

    Confidence: documented
Show secondary and unsupported calls

  • 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.

48.7Qwen3.8-27B55.6Qwen3.8 Max

Like-for-like · BenchAlign v5.7

Qwen3.8 Max leads the like-for-like coding row, although the 90% intervals overlap.

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

Open the coding leaderboardShortlist models for this job
Same basis rules as the category table below

What is actually comparable

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

1 category rests on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.

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 v5.7 lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.

Agentic

Like-for-like
Qwen3.8-27B
61.2
Supported · #16/111
Qwen3.8 Max
64.8
Supported · #11/111
Basis
BenchAlign v5.7 lane · 8 vs 15 public rows
Reading
Qwen3.8 Max leads · intervals overlap

Coding

Like-for-like
Qwen3.8-27B
48.7
Supported · #41/136
Qwen3.8 Max
55.6
Supported · #24/136
Basis
BenchAlign v5.7 lane · 8 vs 12 public rows
Reading
Qwen3.8 Max leads · intervals overlap

Knowledge

Like-for-like
Qwen3.8-27B
49.2
Supported · #59/160
Qwen3.8 Max
66.6
Supported · #19/160
Basis
BenchAlign v5.7 lane · 6 vs 6 public rows
Reading
Qwen3.8 Max leads · intervals overlap

Instruction following

Like-for-like
Qwen3.8-27B
83.2
#45/124
Qwen3.8 Max
90.5
#16/124
Basis
Provisional lane · 1 vs 1 weighted rows
Reading
Qwen3.8 Max leads

Multimodal

Directional only
Qwen3.8-27B
80.9
#11/50
Qwen3.8 Max
88.4
#5/50
Basis
Provisional lane · 1 vs 2 weighted rows
Reading
Directional only

Reasoning

Not comparable
Qwen3.8-27B
78.7
#8/27
Qwen3.8 Max
87.7
Unranked · 2 rankable rows
Basis
Provisional lane · 0 vs 2 weighted rows
Reading
Not comparable

Multilingual

Not comparable
Qwen3.8-27B
Not ranked
Qwen3.8 Max
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Math

Not comparable
Qwen3.8-27B
Not ranked
Qwen3.8 Max
Not ranked
Basis
Provisional lane · 0 vs 0 weighted rows
Reading
Not comparable

Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign v5.7) differ from the provisional-lane categories. Unranked scores sit on the provisional lane’s scale but never name a winner. 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

Qwen3.8-27B
Self-hosted; infrastructure cost varies
Fits in one request
Qwen3.8 Max
API rate not published
Fits in one request

Qwen3.8-27B has no comparable published API token rate. Qwen3.8 Max has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

Qwen3.8-27B
Self-hosted; infrastructure cost varies
Fits in one request
Qwen3.8 Max
API rate not published
Fits in one request

Qwen3.8-27B has no comparable published API token rate. Qwen3.8 Max has no comparable published API token rate.

Cache-heavy agent loop

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

Qwen3.8-27B
Self-hosted; infrastructure cost varies
Fits in one request
Cached-input rate unavailable
Qwen3.8 Max
API rate not published
Fits in one request
Cached-input rate unavailable

Qwen3.8-27B has no comparable published API token rate. 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

Specification differences

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

Documented inputs

Qwen3.8-27B

Not sourced

Qwen3.8 Max

Not sourced

Documented outputs

Qwen3.8-27B

Not sourced

Qwen3.8 Max

Not sourced

Provider availability

Qwen3.8-27B

Not sourced

Qwen3.8 Max

Not sourced

Reasoning profile

Qwen3.8-27B

Reasoning

Qwen3.8 Max

Reasoning

Weight access

Qwen3.8-27B

Open Weight

Qwen3.8 Max

Open Weight

License

Qwen3.8-27B

Open Weight

Qwen3.8 Max

Open Weight

Release date

Qwen3.8-27B

2026-08-05

Qwen3.8 Max

2026-08-03

If you already use one of these models

Deployment change
Both entries list Alibaba as the provider. Confirm endpoint, model ID, limits, and feature support before switching.
Quality signal
Qwen3.8 Max has the higher public score, 71.69 versus 55.26, and the 90% score intervals do not overlap.
Workload cost
A complete comparable API-rate estimate is not available for both models.
Context tradeoff
Qwen3.8 Max has the larger documented window (1M).
Run the same representative tasks against both endpoints before changing production traffic.

Questions

Which is better, Qwen3.8-27B or Qwen3.8 Max?

Qwen3.8 Max has the higher public score, 71.69 versus 55.26, and the 90% score intervals do not overlap. That is the clearest overall quality signal in the current public data.

Which is better for coding, Qwen3.8-27B or Qwen3.8 Max?

Qwen3.8 Max leads the public coding lane, 55.6 to 48.7, with Supported evidence for both models, although the 90% intervals overlap.

Which is better for agentic tasks, Qwen3.8-27B or Qwen3.8 Max?

Qwen3.8 Max leads the public agentic tasks lane, 64.8 to 61.2, with Supported evidence for both models, although the 90% intervals overlap.

Which costs less, Qwen3.8-27B 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, Qwen3.8-27B or Qwen3.8 Max?

Qwen3.8 Max has the larger documented context window: 1M, compared with 262K.

Benchmark evidence

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

Browse raw public benchmark evidence61 rows

Agentic

  • Terminal-Bench 2.1

    Qwen3.8-27B73.0%
    Source
    Qwen3.8 Max86.6%
    Source

    Qwen3.8 Max leads this result

  • CoWorkBench

    Qwen3.8-27B70.7%
    Source
    Qwen3.8 Max74.8%
    Source

    Qwen3.8 Max leads this result

  • JobBench

    Qwen3.8-27B33.4%
    Source
    Qwen3.8 Max53.4%
    Source

    Qwen3.8 Max leads this result

  • Agents' Last Exam

    Qwen3.8-27B42.9%
    Source
    Qwen3.8 Max52.4%
    Source

    Qwen3.8 Max leads this result

  • OSWorld-Verified

    Qwen3.8-27B84.3%
    Source
    Qwen3.8 Max86.1%
    Source

    Qwen3.8 Max leads this result

  • WebArena-Verified

    Qwen3.8-27B64.8%
    Source
    Qwen3.8 Max66.8%
    Source

    Qwen3.8 Max leads this result

  • AndroidWorld

    Qwen3.8-27B81.9%
    Source
    Qwen3.8 Max85.3%
    Source

    Qwen3.8 Max leads this result

  • Terminal-Bench 2.1 (Vals)

    Qwen3.8-27B58.4%
    Source
    Qwen3.8 Max67.4%
    Source

    Qwen3.8 Max leads this result

  • skillsBench

    Qwen3.8-27B—
    Qwen3.8 Max70.2%
    Source

    Not directly comparable

  • AutomationBench

    Qwen3.8-27B—
    Qwen3.8 Max27.3%
    Source

    Not directly comparable

  • Toolathlon-Verified

    Qwen3.8-27B—
    Qwen3.8 Max72.5%
    Source

    Not directly comparable

  • WideResearch

    Qwen3.8-27B—
    Qwen3.8 Max81.9%
    Source

    Not directly comparable

  • HLE w/ tools

    Qwen3.8-27B—
    Qwen3.8 Max56.2%
    Source

    Not directly comparable

  • OSWorld 2.0

    Qwen3.8-27B—
    Qwen3.8 Max19.4%
    Source

    Not directly comparable

  • MobileWorld

    Qwen3.8-27B—
    Qwen3.8 Max77.8%
    Source

    Not directly comparable

Coding

  • Terminal-Bench 2.1

    Qwen3.8-27B73.0%
    Source
    Qwen3.8 Max86.6%
    Source

    Qwen3.8 Max leads this result

  • SWE-bench Pro

    Qwen3.8-27B61.7%
    Source
    Qwen3.8 Max67.7%
    Source

    Qwen3.8 Max leads this result

  • NL2Repo

    Qwen3.8-27B42.3%
    Source
    Qwen3.8 Max55.9%
    Source

    Qwen3.8 Max leads this result

  • DeepSWE

    Qwen3.8-27B42.2%
    Source
    Qwen3.8 Max56.6%
    Source

    Qwen3.8 Max leads this result

  • LiveCodeBench v6

    Qwen3.8-27B90.3%
    Source
    Qwen3.8 Max—

    Not directly comparable

  • VulcanBench v3

    Shared source
    Qwen3.8-27B82.6%
    Qwen3.8 Max81.2%

    Qwen3.8-27B leads this result

  • LiveCodeBench (Vals)

    Qwen3.8-27B84.0%
    Source
    Qwen3.8 Max87.9%
    Source

    Qwen3.8 Max leads this result

  • SWE-bench (Vals)

    Qwen3.8-27B86.0%
    Source
    Qwen3.8 Max85.6%
    Source

    Qwen3.8-27B leads this result

  • FrontierSWE

    Qwen3.8-27B—
    Qwen3.8 Max73.5%
    Source

    Not directly comparable

  • MLS-Bench Lite

    Qwen3.8-27B—
    Qwen3.8 Max41.0%
    Source

    Not directly comparable

  • PaperBench

    Qwen3.8-27B—
    Qwen3.8 Max93.0%
    Source

    Not directly comparable

  • OpenHarmony Bench

    Qwen3.8-27B—
    Qwen3.8 Max60.8%
    Source

    Not directly comparable

  • FrontierSWE v2

    Qwen3.8-27B—
    Qwen3.8 Max15.8%
    Source

    Not directly comparable

Reasoning

  • MRCRv2

    Qwen3.8-27B—
    Qwen3.8 Max92.9%
    Source

    Not directly comparable

  • LongBench v2

    Qwen3.8-27B—
    Qwen3.8 Max66.3%
    Source

    Not directly comparable

Multimodal

  • MathVision

    Qwen3.8-27B90.0%
    Source
    Qwen3.8 Max95.2%
    Source

    Qwen3.8 Max leads this result

  • MathVision w/ Python

    Qwen3.8-27B94.6%
    Source
    Qwen3.8 Max97.7%
    Source

    Qwen3.8 Max leads this result

  • BabyVision

    Qwen3.8-27B65.7%
    Source
    Qwen3.8 Max82.0%
    Source

    Qwen3.8 Max leads this result

  • BabyVision w/ Python

    Qwen3.8-27B85.6%
    Source
    Qwen3.8 Max91.3%
    Source

    Qwen3.8 Max leads this result

  • Vision2Web

    Qwen3.8-27B62.9%
    Source
    Qwen3.8 Max69.0%
    Source

    Qwen3.8 Max leads this result

  • CharXiv w/o tools

    Qwen3.8-27B83.7%
    Source
    Qwen3.8 Max88.4%
    Source

    Qwen3.8 Max leads this result

  • CharXiv

    Qwen3.8-27B90.2%
    Source
    Qwen3.8 Max93.5%
    Source

    Qwen3.8 Max leads this result

  • OmniDocBench 1.5

    Qwen3.8-27B91.1%
    Source
    Qwen3.8 Max92.1%
    Source

    Qwen3.8 Max leads this result

  • RealWorldQA

    Qwen3.8-27B85.9%
    Source
    Qwen3.8 Max88.0%
    Source

    Qwen3.8 Max leads this result

  • ERQA

    Qwen3.8-27B65.5%
    Source
    Qwen3.8 Max77.8%
    Source

    Qwen3.8 Max leads this result

  • MMMU-Pro

    Qwen3.8-27B—
    Qwen3.8 Max82.3%
    Source

    Not directly comparable

  • ZeroBench

    Qwen3.8-27B—
    Qwen3.8 Max24.0%
    Source

    Not directly comparable

  • ZeroBench w/ Python

    Qwen3.8-27B—
    Qwen3.8 Max49.0%
    Source

    Not directly comparable

  • MedXpertQA (MM)

    Qwen3.8-27B—
    Qwen3.8 Max80.4%
    Source

    Not directly comparable

  • ScreenSpot Pro

    Qwen3.8-27B—
    Qwen3.8 Max84.5%
    Source

    Not directly comparable

  • OCRBench V2

    Qwen3.8-27B—
    Qwen3.8 Max74.2%
    Source

    Not directly comparable

  • CC-OCR

    Qwen3.8-27B—
    Qwen3.8 Max79.6%
    Source

    Not directly comparable

  • SimpleVQA

    Qwen3.8-27B—
    Qwen3.8 Max75.0%
    Source

    Not directly comparable

  • PerceptionBench

    Qwen3.8-27B—
    Qwen3.8 Max63.5%
    Source

    Not directly comparable

  • Video-MME (with subtitle)

    Qwen3.8-27B—
    Qwen3.8 Max90.4%
    Source

    Not directly comparable

  • VideoMMMU

    Qwen3.8-27B—
    Qwen3.8 Max88.7%
    Source

    Not directly comparable

  • MMVU

    Qwen3.8-27B—
    Qwen3.8 Max82.4%
    Source

    Not directly comparable

  • MLVU (M-Avg)

    Qwen3.8-27B—
    Qwen3.8 Max90.8%
    Source

    Not directly comparable

  • LVBench

    Qwen3.8-27B—
    Qwen3.8 Max81.8%
    Source

    Not directly comparable

Knowledge

  • GPQA

    Qwen3.8-27B89.2%
    Source
    Qwen3.8 Max92.6%
    Source

    Qwen3.8 Max leads this result

  • GPQA-D

    Qwen3.8-27B89.2%
    Source
    Qwen3.8 Max92.6%
    Source

    Qwen3.8 Max leads this result

  • HLE

    Qwen3.8-27B30.8%
    Source
    Qwen3.8 Max43.6%
    Source

    Qwen3.8 Max leads this result

  • HLE w/o tools

    Qwen3.8-27B30.8%
    Source
    Qwen3.8 Max43.6%
    Source

    Qwen3.8 Max leads this result

  • GPQA Diamond (Vals)

    Qwen3.8-27B88.9%
    Source
    Qwen3.8 Max93.7%
    Source

    Qwen3.8 Max leads this result

  • MMLU-Pro (Vals)

    Qwen3.8-27B84.3%
    Source
    Qwen3.8 Max88.6%
    Source

    Qwen3.8 Max leads this result

Instruction following

  • IFBench

    Qwen3.8-27B79.5%
    Source
    Qwen3.8 Max82.8%
    Source

    Qwen3.8 Max leads this result

61 public results · 32 shared

Related comparisons

Compare API pricingRead the methodologyOpen the model selector

Watch Qwen3.8-27B 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 2,000+ readers.

Last updated September 28, 2026