Skip to main content
Radar

Five or fewer confirmed AI changes, with original sources, on mornings when something changed.A free source-linked morning brief.

Start free brief
BenchLM
Model A
Qwen3.5-27B

Alibaba

59.6/100

Supported · Public rank #58

90% interval 50.4–68.9

Qwen3.5-27B vs Qwen3.5-35B-A3B

Updated August 13, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.

Model B
Qwen3.5-35B-A3B

Alibaba

56.0/100

Supported · Public rank #86

90% interval 44.6–67.4

Decision reading

Qwen3.5-27B has the higher public score estimate, 59.64 versus 55.97, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.

16 results are shared. Category rows based on different benchmark sets are marked directional and do not name a winner.

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

    Qwen3.5-27B

    Qwen3.5-27B leads on the same 2 weighted benchmark rows.

    Confidence: limited

  • Agentic work

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

    Qwen3.5-27B

    Qwen3.5-27B leads on the same 3 weighted benchmark rows.

    Confidence: stronger

Show secondary and unsupported calls

  • 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

What is actually comparable

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

Shared results
16
Qwen3.5-27B only
0
Qwen3.5-35B-A3B only
0
Like-for-like categories
6 / 8

Category results, on a stated basis

Each row states whether both averages use the same weighted benchmark set. Directional and not-comparable rows remain visible, but they never receive a winner in this template.

Agentic

Like-for-like
Qwen3.5-27B
52.0
Qwen3.5-35B-A3B
51.0
Weighted basis
3 vs 3 rows
Reading
Qwen3.5-27B leads

Coding

Like-for-like
Qwen3.5-27B
64.9
Qwen3.5-35B-A3B
60.6
Weighted basis
2 vs 2 rows
Reading
Qwen3.5-27B leads

Reasoning

Like-for-like
Qwen3.5-27B
60.6
Qwen3.5-35B-A3B
59.0
Weighted basis
1 vs 1 rows
Reading
Qwen3.5-27B leads

Knowledge

Like-for-like
Qwen3.5-27B
82.7
Qwen3.5-35B-A3B
81.6
Weighted basis
3 vs 3 rows
Reading
Qwen3.5-27B leads

Multilingual

Like-for-like
Qwen3.5-27B
82.2
Qwen3.5-35B-A3B
81.0
Weighted basis
1 vs 1 rows
Reading
Qwen3.5-27B leads

Instruction following

Like-for-like
Qwen3.5-27B
95.0
Qwen3.5-35B-A3B
91.9
Weighted basis
1 vs 1 rows
Reading
Qwen3.5-27B leads

Math

Not comparable
Qwen3.5-27B
Not measured
Qwen3.5-35B-A3B
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Multimodal

Not comparable
Qwen3.5-27B
Not measured
Qwen3.5-35B-A3B
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

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.

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

Qwen3.5-27B
Self-hosted; infrastructure cost varies
Fits in one request
Qwen3.5-35B-A3B
Self-hosted; infrastructure cost varies
Fits in one request

Qwen3.5-27B has no comparable published API token rate. Qwen3.5-35B-A3B has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

Qwen3.5-27B
Self-hosted; infrastructure cost varies
Fits in one request
Qwen3.5-35B-A3B
Self-hosted; infrastructure cost varies
Fits in one request

Qwen3.5-27B has no comparable published API token rate. Qwen3.5-35B-A3B has no comparable published API token rate.

Cache-heavy agent loop

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

Qwen3.5-27B
Self-hosted; infrastructure cost varies
Fits in one request
Cached-input rate unavailable
Qwen3.5-35B-A3B
Self-hosted; infrastructure cost varies
Fits in one request
Cached-input rate unavailable

Qwen3.5-27B has no comparable published API token rate. Qwen3.5-35B-A3B has no comparable published API token rate.

Specification differences

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

Context window

Maximum documented context; output-token limits may be lower.

Qwen3.5-27B

262K

Qwen3.5-35B-A3B

262K

API model ID

Qwen3.5-27B

Not sourced

Qwen3.5-35B-A3B

Not sourced

Cached-input rate

A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.

Qwen3.5-27B

No comparable hosted API rate

Qwen3.5-35B-A3B

No comparable hosted API rate

Documented inputs

Qwen3.5-27B

Not sourced

Qwen3.5-35B-A3B

Not sourced

Documented outputs

Qwen3.5-27B

Not sourced

Qwen3.5-35B-A3B

Not sourced

Provider availability

Qwen3.5-27B

Not sourced

Qwen3.5-35B-A3B

Not sourced

Reasoning profile

Qwen3.5-27B

Reasoning

Qwen3.5-35B-A3B

Reasoning

Weight access

Qwen3.5-27B

Open Weight

Qwen3.5-35B-A3B

Open Weight

License

Qwen3.5-27B

Open Weight

Qwen3.5-35B-A3B

Open Weight

Release date

Qwen3.5-27B

2026-03-04

Qwen3.5-35B-A3B

2026-03-04

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
Qwen3.5-27B has the higher public score estimate, 59.64 versus 55.97, 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 262K.

Run the same representative tasks against both endpoints before changing production traffic.

Benchmark evidence

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

Browse raw public benchmark evidence16 rows

Agentic

  • Terminal-Bench 2.0

    Qwen3.5-27B41.6%
    Source
    Qwen3.5-35B-A3B40.5%
    Source

    Qwen3.5-27B leads this result

  • BrowseComp

    Qwen3.5-27B61%
    Source
    Qwen3.5-35B-A3B61%
    Source

    Tie

  • OSWorld-Verified

    Qwen3.5-27B56.2%
    Source
    Qwen3.5-35B-A3B54.5%
    Source

    Qwen3.5-27B leads this result

  • Gert Labs

    Shared source
    Qwen3.5-27B39.41%
    Qwen3.5-35B-A3B28.96%

    Qwen3.5-27B leads this result

Coding

  • SWE-bench Verified

    Qwen3.5-27B72.4%
    Source
    Qwen3.5-35B-A3B69.2%
    Source

    Qwen3.5-27B leads this result

  • SWE-Rebench

    Shared source
    Qwen3.5-27B58.9%
    Qwen3.5-35B-A3B53.7%

    Qwen3.5-27B leads this result

Reasoning

  • LongBench v2

    Qwen3.5-27B60.6%
    Source
    Qwen3.5-35B-A3B59%
    Source

    Qwen3.5-27B leads this result

Knowledge

  • MMLU-Pro

    Qwen3.5-27B86.1%
    Source
    Qwen3.5-35B-A3B85.3%
    Source

    Qwen3.5-27B leads this result

  • SuperGPQA

    Qwen3.5-27B65.6%
    Source
    Qwen3.5-35B-A3B63.4%
    Source

    Qwen3.5-27B leads this result

  • GPQA

    Qwen3.5-27B85.5%
    Source
    Qwen3.5-35B-A3B84.2%
    Source

    Qwen3.5-27B leads this result

Multilingual

  • MMLU-ProX

    Qwen3.5-27B82.2%
    Source
    Qwen3.5-35B-A3B81%
    Source

    Qwen3.5-27B leads this result

Multimodal

  • MMMU

    Shared source
    Qwen3.5-27B82.3%
    Qwen3.5-35B-A3B81.4%

    Qwen3.5-27B leads this result

  • MMVU

    Shared source
    Qwen3.5-27B73.3%
    Qwen3.5-35B-A3B72.3%

    Qwen3.5-27B leads this result

  • MathVision

    Shared source
    Qwen3.5-27B86.0%
    Qwen3.5-35B-A3B83.9%

    Qwen3.5-27B leads this result

  • V*

    Shared source
    Qwen3.5-27B93.7%
    Qwen3.5-35B-A3B92.7%

    Qwen3.5-27B leads this result

Instruction following

  • IFEval

    Qwen3.5-27B95%
    Source
    Qwen3.5-35B-A3B91.9%
    Source

    Qwen3.5-27B leads this result

Frequently asked questions

Which is better, Qwen3.5-27B or Qwen3.5-35B-A3B?

Qwen3.5-27B has the higher public score estimate, 59.64 versus 55.97, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.

Which is better for coding, Qwen3.5-27B or Qwen3.5-35B-A3B?

Qwen3.5-27B leads the like-for-like coding comparison across 2 shared weighted benchmark rows.

Which is better for agentic tasks, Qwen3.5-27B or Qwen3.5-35B-A3B?

Qwen3.5-27B leads the like-for-like agentic tasks comparison across 3 shared weighted benchmark rows.

Which costs less, Qwen3.5-27B or Qwen3.5-35B-A3B?

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, Qwen3.5-27B or Qwen3.5-35B-A3B?

Both models list the same context window, 262K.

Related comparisons

Compare API pricing Read the methodology Open the model selector

Last updated August 13, 2026

Watch Qwen3.5-27B vs Qwen3.5-35B-A3B

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

Read a sample issue

Join 2,000+ readers.