Model A

Alibaba

56.44/100

Estimated · Public rank #97

90% interval 44.9–68.0

Qwen3.5 397B vs Qwen3.6-27B

Updated September 4, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.

Model B

Qwen3.6-27B

Alibaba

52.73/100

Estimated · Public rank #120

90% interval 47.0–58.5

Decision reading

Qwen3.5 397B has the higher public score estimate, 56.44 versus 52.73, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.

21 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Share or export

Share on X LinkedInSocial card CSV JSON

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.

Long documents
Prompts that approach the documented context limit
Qwen3.6-27B
Qwen3.6-27B has the larger documented context window.
Confidence: documented

Show secondary and unsupported calls

Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
Qwen3.5 397B is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
Qwen3.5 397B is scored on Estimated evidence for agentic, so the reading is directional rather than like-for-like.
Confidence: limited
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
The page does not recommend a cost winner because at least one model cannot fit the stated workload in one request. Qwen3.5 397B does not fit this workload in one request. Qwen3.5 397B has no published cached-input rate, so cached tokens use its listed input rate. Qwen3.6-27B has no comparable published API token rate.
Confidence: rate-fallback
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.

17 Qwen3.5 397B only

21 Shared

17 Qwen3.6-27B only

Shared results: 21
Qwen3.5 397B only: 17
Qwen3.6-27B only: 17
Like-for-like categories: 1 / 8

4 categories rest on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.

Category results, on a stated basis

Each row shows the public-lane category score for both models: the BenchAlign 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.

Multimodal

Like-for-like

Qwen3.5 397B: 62.2; #27/48
Qwen3.6-27B: 51.5; #35/48
Basis: Provisional lane · 2 vs 2 weighted rows
Reading: Qwen3.5 397B leads

Agentic

Directional only

Qwen3.5 397B: 48.2; Estimated · #78/151
Qwen3.6-27B: 33.9; Supported · #133/151
Basis: BenchAlign lane · 13 vs 6 public rows
Reading: Directional only

Coding

Directional only

Qwen3.5 397B: 52.4; Estimated · #56/183
Qwen3.6-27B: 42.6; Supported · #128/183
Basis: BenchAlign lane · 3 vs 6 public rows
Reading: Directional only

Knowledge

Directional only

Qwen3.5 397B: 51.5; Estimated · #82/181
Qwen3.6-27B: 49.0; Estimated · #95/181
Basis: BenchAlign lane · 6 vs 6 public rows
Reading: Directional only

Instruction following

Directional only

Qwen3.5 397B: 0.0; #120/120
Qwen3.6-27B: 82.2; #50/120
Basis: Provisional lane · 0 vs 0 weighted rows
Reading: Directional only

Reasoning

Not comparable

Qwen3.5 397B: 59.8; Unranked · 2 rankable rows
Qwen3.6-27B: 73.7; Unranked · 2 rankable rows
Basis: Provisional lane · 1 vs 0 weighted rows
Reading: Not comparable

Math

Not comparable

Qwen3.5 397B: 74.2; Unranked · 5 rankable rows
Qwen3.6-27B: 72.8; Unranked · 5 rankable rows
Basis: Provisional lane · 2 vs 2 weighted rows
Reading: Not comparable

Multilingual

Not comparable

Qwen3.5 397B: 69.7; #5/12
Qwen3.6-27B: Not ranked
Basis: Provisional lane · 1 vs 0 weighted rows
Reading: Not comparable

Category scores with the server-provided evidence basis for Qwen3.5 397B and Qwen3.6-27B
Category	Qwen3.5 397B	Qwen3.6-27B	Basis	Reading
Multimodal	62.2#27/48	51.5#35/48	Like-for-likeProvisional lane · 2 vs 2 weighted rows	Qwen3.5 397B leads
Agentic	48.2Estimated · #78/151	33.9Supported · #133/151	Directional onlyBenchAlign lane · 13 vs 6 public rows	Directional only
Coding	52.4Estimated · #56/183	42.6Supported · #128/183	Directional onlyBenchAlign lane · 3 vs 6 public rows	Directional only
Knowledge	51.5Estimated · #82/181	49.0Estimated · #95/181	Directional onlyBenchAlign lane · 6 vs 6 public rows	Directional only
Instruction following	0.0#120/120	82.2#50/120	Directional onlyProvisional lane · 0 vs 0 weighted rows	Directional only
Reasoning	59.8Unranked · 2 rankable rows	73.7Unranked · 2 rankable rows	Not comparableProvisional lane · 1 vs 0 weighted rows	Not comparable
Math	74.2Unranked · 5 rankable rows	72.8Unranked · 5 rankable rows	Not comparableProvisional lane · 2 vs 2 weighted rows	Not comparable
Multilingual	69.7#5/12	Not ranked	Not comparableProvisional lane · 1 vs 0 weighted rows	Not comparable

Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign) differ from the provisional-lane categories. Unranked scores sit on the lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.

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.

Terminal-Bench 2.0
Agentic
Qwen3.5 397B: 52.5%Qwen3.6-27B: 59.3%Normalized gap 6.8Qwen3.5 397B source Qwen3.6-27B source
HLE
Knowledge
Qwen3.5 397B: 28.7%Qwen3.6-27B: 24%Normalized gap 4.7Qwen3.5 397B source Qwen3.6-27B source
SuperGPQA
Knowledge
Qwen3.5 397B: 70.4%Qwen3.6-27B: 66%Normalized gap 4.4Qwen3.5 397B source Qwen3.6-27B source
HMMT Feb 2026
Math
Qwen3.5 397B: 87.9%Qwen3.6-27B: 84.3%Normalized gap 3.6Qwen3.5 397B source Qwen3.6-27B source
MMMU-Pro
Multimodal
Qwen3.5 397B: 79%Qwen3.6-27B: 75.8%Normalized gap 3.2Qwen3.5 397B source Qwen3.6-27B source

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 397B: $0.0024; Fits in one request
Qwen3.6-27B: Self-hosted; infrastructure cost varies; Fits in one request

Qwen3.6-27B has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

Qwen3.5 397B: $0.0408; Fits in one request
Qwen3.6-27B: Self-hosted; infrastructure cost varies; Fits in one request

Qwen3.6-27B has no comparable published API token rate.

Cache-heavy agent loop

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

Qwen3.5 397B: $0.168; Does not fit in one request; Cached input priced at the published list-input rate
Qwen3.6-27B: Self-hosted; infrastructure cost varies; Fits in one request; Cached-input rate unavailable

Qwen3.5 397B does not fit this workload in one request. Qwen3.5 397B has no published cached-input rate, so cached tokens use its listed input rate. Qwen3.6-27B 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.

SpecificationQwen3.5 397BQwen3.6-27B

Context window

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

Qwen3.5 397B

128K

Qwen3.6-27B

262K

API model ID

Qwen3.5 397B

Not sourced

Qwen3.6-27B

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 397B

Not published

Qwen3.6-27B

No comparable hosted API rate

Documented inputs

Qwen3.5 397B

Not sourced

Qwen3.6-27B

Not sourced

Documented outputs

Qwen3.5 397B

Not sourced

Qwen3.6-27B

Not sourced

Provider availability

Qwen3.5 397B

Not sourced

Qwen3.6-27B

Not sourced

Reasoning profile

Qwen3.5 397B

Non-Reasoning

Qwen3.6-27B

Reasoning

Weight access

Qwen3.5 397B

Open Weight

Qwen3.6-27B

Open Weight

License

Qwen3.5 397B

Open Weight

Qwen3.6-27B

Open Weight

Release date

Qwen3.5 397B

2026-02-16

Qwen3.6-27B

2026-04-21

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 397B has the higher public score estimate, 56.44 versus 52.73, but the 90% score intervals overlap.
Workload cost: A complete comparable API-rate estimate is not available for both models.
Context tradeoff: Qwen3.6-27B has the larger documented window (262K).

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

Self-host vs API cost

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

Qwen3.5 397B

API / mo$3,150

Self-host / moNot listed

Break-even—

Proprietary model — self-hosting not applicable.

Qwen3.6-27B

API / mo$0

Self-host / mo$429

Break-even—

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 evidence55 rows

Agentic

Terminal-Bench 2.0
Qwen3.5 397B52.5%
Source
Qwen3.6-27B59.3%
Source
Qwen3.6-27B leads this result
BrowseComp
Qwen3.5 397B62%
Source
Qwen3.6-27B—
Not directly comparable
Claw-Eval
Qwen3.5 397B56.8%
Source
Qwen3.6-27B72.4%
Source
Qwen3.6-27B leads this result
QwenClawBench
Qwen3.5 397B51.8%
Source
Qwen3.6-27B53.4%
Source
Qwen3.6-27B leads this result
τ³-bench results
Qwen3.5 397B68.4%
Source
Qwen3.6-27B—
Not directly comparable
VITA-Bench
Qwen3.5 397B43.7%
Source
Qwen3.6-27B—
Not directly comparable
DeepPlanning
Qwen3.5 397B37.6%
Source
Qwen3.6-27B—
Not directly comparable
Toolathlon
Qwen3.5 397B36.3%
Source
Qwen3.6-27B—
Not directly comparable
MCP Atlas
Qwen3.5 397B46.1%
Source
Qwen3.6-27B—
Not directly comparable
MCP-Tasks
Qwen3.5 397B74.2%
Source
Qwen3.6-27B—
Not directly comparable
WideResearch
Qwen3.5 397B74.0%
Source
Qwen3.6-27B—
Not directly comparable
Gert Labs
Shared source
Qwen3.5 397B46.76%
Qwen3.6-27B54.84%
Qwen3.6-27B leads this result
ResearchClawBench
Qwen3.5 397B14.2%
Source
Qwen3.6-27B—
Not directly comparable
QwenWebBench
Qwen3.5 397B—
Qwen3.6-27B1487
Source
Not directly comparable
AndroidWorld
Qwen3.5 397B—
Qwen3.6-27B70.3%
Source
Not directly comparable

Coding

SWE-bench Verified
Qwen3.5 397B76.2%
Source
Qwen3.6-27B77.2%
Source
Qwen3.6-27B leads this result
LiveCodeBench v6
Qwen3.5 397B83.6%
Source
Qwen3.6-27B—
Not directly comparable
SWE-bench Pro
Qwen3.5 397B50.9%
Source
Qwen3.6-27B53.5%
Source
Qwen3.6-27B leads this result
SWE Multilingual
Qwen3.5 397B—
Qwen3.6-27B71.3%
Source
Not directly comparable
Terminal-Bench 2.0
Qwen3.5 397B—
Qwen3.6-27B59.3%
Source
Not directly comparable
LiveCodeBench
Qwen3.5 397B—
Qwen3.6-27B83.9%
Source
Not directly comparable
NL2Repo
Qwen3.5 397B—
Qwen3.6-27B36.2%
Source
Not directly comparable

Reasoning

LongBench v2
Qwen3.5 397B63.2%
Source
Qwen3.6-27B—
Not directly comparable
AI-Needle
Qwen3.5 397B68.7%
Source
Qwen3.6-27B—
Not directly comparable

Knowledge

GPQA
Qwen3.5 397B88.4%
Source
Qwen3.6-27B87.8%
Source
Qwen3.5 397B leads this result
SuperGPQA
Qwen3.5 397B70.4%
Source
Qwen3.6-27B66%
Source
Qwen3.5 397B leads this result
MMLU-Pro
Qwen3.5 397B87.8%
Source
Qwen3.6-27B86.2%
Source
Qwen3.5 397B leads this result
MMLU-Redux
Qwen3.5 397B94.9%
Source
Qwen3.6-27B93.5%
Source
Qwen3.5 397B leads this result
C-Eval
Qwen3.5 397B93%
Source
Qwen3.6-27B91.4%
Source
Qwen3.5 397B leads this result
HLE
Qwen3.5 397B28.7%
Source
Qwen3.6-27B24%
Source
Qwen3.5 397B leads this result

Math

AIME26
Qwen3.5 397B93.3%
Source
Qwen3.6-27B94.1%
Source
Qwen3.6-27B leads this result
HMMT Feb 2025
Qwen3.5 397B94.8%
Source
Qwen3.6-27B93.8%
Source
Qwen3.5 397B leads this result
HMMT Nov 2025
Qwen3.5 397B92.7%
Source
Qwen3.6-27B90.7%
Source
Qwen3.5 397B leads this result
HMMT Feb 2026
Qwen3.5 397B87.9%
Source
Qwen3.6-27B84.3%
Source
Qwen3.5 397B leads this result
MMAnswerBench
Qwen3.5 397B80.9%
Source
Qwen3.6-27B80.8%
Source
Qwen3.5 397B leads this result

Multilingual

MMLU-ProX
Qwen3.5 397B84.7%
Source
Qwen3.6-27B—
Not directly comparable
NOVA-63
Qwen3.5 397B59.1%
Source
Qwen3.6-27B—
Not directly comparable

Multimodal

MMMU-Pro
Qwen3.5 397B79%
Source
Qwen3.6-27B75.8%
Source
Qwen3.5 397B leads this result
MathVision
Qwen3.5 397B88.6%
Source
Qwen3.6-27B—
Not directly comparable
CharXiv
Qwen3.5 397B80.8%
Source
Qwen3.6-27B78.4%
Source
Qwen3.5 397B leads this result
VideoMMMU
Qwen3.5 397B84.7%
Source
Qwen3.6-27B84.4%
Source
Qwen3.5 397B leads this result
ScreenSpot Pro
Qwen3.5 397B65.6%
Source
Qwen3.6-27B—
Not directly comparable
V*
Qwen3.5 397B95.8%
Source
Qwen3.6-27B94.7%
Source
Qwen3.5 397B leads this result
MMMU
Qwen3.5 397B—
Qwen3.6-27B82.9%
Source
Not directly comparable
RealWorldQA
Qwen3.5 397B—
Qwen3.6-27B84.1%
Source
Not directly comparable
DynaMath
Qwen3.5 397B—
Qwen3.6-27B85.6%
Source
Not directly comparable
MStar
Qwen3.5 397B—
Qwen3.6-27B81.4%
Source
Not directly comparable
SimpleVQA
Qwen3.5 397B—
Qwen3.6-27B56.1%
Source
Not directly comparable
CC-OCR
Qwen3.5 397B—
Qwen3.6-27B81.2%
Source
Not directly comparable
CountBench
Qwen3.5 397B—
Qwen3.6-27B97.8%
Source
Not directly comparable
RefCOCO (avg)
Qwen3.5 397B—
Qwen3.6-27B92.5%
Source
Not directly comparable
ERQA
Qwen3.5 397B—
Qwen3.6-27B62.5%
Source
Not directly comparable
Video-MME (with subtitle)
Qwen3.5 397B—
Qwen3.6-27B87.7%
Source
Not directly comparable
MLVU (M-Avg)
Qwen3.5 397B—
Qwen3.6-27B86.6%
Source
Not directly comparable

Instruction following

IFEval
Qwen3.5 397B92.6%
Source
Qwen3.6-27B—
Not directly comparable

Frequently asked questions

Which is better, Qwen3.5 397B or Qwen3.6-27B?

Qwen3.5 397B has the higher public score estimate, 56.44 versus 52.73, 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 397B or Qwen3.6-27B?

Qwen3.5 397B scores higher for coding on the public lane, 52.4 to 42.6. Qwen3.5 397B is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like. BenchLM does not name a winner for coding from a directional reading; read the shared benchmark rows directly and test the models on the same task set.

Which is better for agentic tasks, Qwen3.5 397B or Qwen3.6-27B?

Qwen3.5 397B scores higher for agentic tasks on the public lane, 48.2 to 33.9. Qwen3.5 397B is scored on Estimated evidence for agentic tasks, so the reading is directional rather than like-for-like. BenchLM does not name a winner for agentic tasks from a directional reading; read the shared benchmark rows directly and test the models on the same task set.

Which costs less, Qwen3.5 397B or Qwen3.6-27B?

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 397B or Qwen3.6-27B?

Qwen3.6-27B has the larger documented context window: 262K, compared with 128K.

Related comparisons

Compare API pricing Read the methodology Open the model selector

Last updated September 4, 2026

Watch Qwen3.5 397B vs Qwen3.6-27B

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

Read a sample issue

Join 2,000+ readers.

Qwen3.5 397B vs Qwen3.6-27B

Which one for your work

Long documents

Coding work

Agentic work

Chat turn cost

Cache-heavy agent loop cost

Repository review cost

What is actually comparable

Category results, on a stated basis

Multimodal

Agentic

Coding

Knowledge

Instruction following

Reasoning

Math

Multilingual

Shape of the matched evidence

What each workload costs

Chat turn

Repository review

Cache-heavy agent loop

Specification differences

Context window

API model ID

Cached-input rate

Documented inputs

Documented outputs

Provider availability

Reasoning profile

Weight access

License

Release date

Self-host vs API cost

Benchmark evidence

Agentic

Coding

Reasoning

Knowledge

Math

Multilingual

Multimodal

Instruction following

Frequently asked questions

Which is better, Qwen3.5 397B or Qwen3.6-27B?

Which is better for coding, Qwen3.5 397B or Qwen3.6-27B?

Which is better for agentic tasks, Qwen3.5 397B or Qwen3.6-27B?

Which costs less, Qwen3.5 397B or Qwen3.6-27B?

Which has the larger context window, Qwen3.5 397B or Qwen3.6-27B?

Related comparisons

Watch Qwen3.5 397B vs Qwen3.6-27B