Skip to main content

Model comparison

Qwen3.5 397B vs Qwen3.6-35B-A3B

Data verified

Head-to-head evidence from 42 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

56.73/100
Margin
5.4pts
← winning
51.36/100
4 category wins1 category wins

Verified leaderboard positions: Qwen3.5 397B #20; Qwen3.6-35B-A3B #31

BenchAlign evidence: Qwen3.5 397B estimated; Qwen3.6-35B-A3B estimated. Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Qwen3.5 397B and Qwen3.6-35B-A3B share 42 comparable benchmark results. 5 of 8 categories are comparable. 14 results are unique to Qwen3.5 397B; 16 to Qwen3.6-35B-A3B.

Updated July 15, 2026
Shared results
42
Qwen3.5 397B only
14
Qwen3.6-35B-A3B only
16
Comparable categories
5 / 8

Treat this as a split decision. Qwen3.5 397B makes more sense if knowledge is the priority or you would rather avoid the extra latency and token burn of a reasoning model; Qwen3.6-35B-A3B is the better fit if coding is the priority or you need the larger 262K context window.

Confidence note. This is a partial-evidence comparison with 42 shared benchmark results across 7 evidence categories; 5 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

Qwen3.5 397B and Qwen3.6-35B-A3B finish on the same provisional overall score, so this is less about a single winner and more about where the edge shows up. The provisional headline says tie; the benchmark table is where the real choice happens.

Qwen3.6-35B-A3B is the reasoning model in the pair, while Qwen3.5 397B is not. That usually helps on harder chain-of-thought-heavy tests, but it can also mean more latency and more token spend in real use. Qwen3.6-35B-A3B gives you the larger context window at 262K, compared with 128K for Qwen3.5 397B.

Category breakdown

Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.

Category scores and score margins for Qwen3.5 397B and Qwen3.6-35B-A3B
CategoryQwen3.5 397BΔQwen3.6-35B-A3B
CodingQwen3.5 397B66.5Margin 7.3Qwen3.6-35B-A3B73.8
KnowledgeQwen3.5 397B56.9Margin 5.1Qwen3.6-35B-A3B51.8
AgenticQwen3.5 397B56.5Margin 5.0Qwen3.6-35B-A3B51.5
MultimodalQwen3.5 397B79.6Margin 3.3Qwen3.6-35B-A3B76.3
MathQwen3.5 397B90.6Margin 2.4Qwen3.6-35B-A3B88.2
ReasoningQwen3.5 397B63.2MarginNo overlapQwen3.6-35B-A3BNot measured
MultilingualQwen3.5 397B84.7MarginNo overlapQwen3.6-35B-A3BNot measured
Inst. FollowingQwen3.5 397B92.6MarginNo overlapQwen3.6-35B-A3BNot measured

Decisive benchmark drivers

The largest measured benchmark gaps in this matchup, with exact reported values.

More
A · Qwen3.5 397BB · Qwen3.6-35B-A3B
  1. HLE

    Knowledge
    Source ↗
    A 28.7%B 21.4%
    Winner: Qwen3.5 397BΔ 7.3
    HLE: Qwen3.5 397B scored 28.7%; Qwen3.6-35B-A3B scored 21.4%. Qwen3.5 397B wins this benchmark.
  2. SuperGPQA

    Knowledge
    Source ↗
    A 70.4%B 64.7%
    Winner: Qwen3.5 397BΔ 5.7
    SuperGPQA: Qwen3.5 397B scored 70.4%; Qwen3.6-35B-A3B scored 64.7%. Qwen3.5 397B wins this benchmark.
  3. HMMT Feb 2026

    Math
    Source ↗
    A 87.9%B 83.6%
    Winner: Qwen3.5 397BΔ 4.3
    HMMT Feb 2026: Qwen3.5 397B scored 87.9%; Qwen3.6-35B-A3B scored 83.6%. Qwen3.5 397B wins this benchmark.
  4. MMMU-Pro

    Multimodal
    Source ↗
    A 79%B 75.3%
    Winner: Qwen3.5 397BΔ 3.7
    MMMU-Pro: Qwen3.5 397B scored 79%; Qwen3.6-35B-A3B scored 75.3%. Qwen3.5 397B wins this benchmark.
  5. SWE-bench Verified

    Coding
    Source ↗
    A 76.2%B 73.4%
    Winner: Qwen3.5 397BΔ 2.8
    SWE-bench Verified: Qwen3.5 397B scored 76.2%; Qwen3.6-35B-A3B scored 73.4%. Qwen3.5 397B wins this benchmark.

Operational comparison

Runtime and commercial metrics are compared only when both models have a complete sourced value.

MetricQwen3.5 397BQwen3.6-35B-A3BComparison
Input / output priceUSD per 1M tokensQwen3.5 397B$0.6 input / $3.6 outputQwen3.6-35B-A3BNot availableA complete price comparison is not available.
Generation speedtokens per secondQwen3.5 397B96 tok/sQwen3.6-35B-A3BNot availableA complete speed comparison is not available.
First-answer latencyseconds to first tokenQwen3.5 397B2.44 sQwen3.6-35B-A3BNot availableA complete latency comparison is not available.
Context windowmaximum listed tokensQwen3.5 397B128KQwen3.6-35B-A3B262KQwen3.6-35B-A3B lists the larger context window.

Benchmark Deep Dive

AgenticQwen3.5 397B wins
BenchmarkQwen3.5 397BQwen3.6-35B-A3BResult
Terminal-Bench 2.0Source 52.5%51.5%Qwen3.5 397B leads
BrowseCompSource 62%Not comparable
Claw-EvalSource 56.8%68.7%Qwen3.6-35B-A3B leads
QwenClawBenchSource 51.8%52.6%Qwen3.6-35B-A3B leads
τ³-bench resultsSource 68.4%67.2%Qwen3.5 397B leads
VITA-BenchSource 43.7%35.6%Qwen3.5 397B leads
DeepPlanningSource 37.6%25.9%Qwen3.5 397B leads
ToolathlonSource 36.3%26.9%Qwen3.5 397B leads
MCP AtlasSource 46.1%62.8%Qwen3.6-35B-A3B leads
MCP-TasksSource 74.2%Not comparable
WideResearchSource 74.0%60.1%Qwen3.5 397B leads
τ²-bench resultsSource 95.6%95.3%Qwen3.5 397B leads
Gert LabsSource 46.76%42.65%Qwen3.5 397B leads
ResearchClawBenchSource 14.2%Not comparable
AA Agentic IndexSource 19.9%21.4%Qwen3.6-35B-A3B leads
APEX-Agents-AASource 15.3%Not comparable
GDPval-AASource 23.1%27.4%Qwen3.6-35B-A3B leads
GDPval-AASource 9621049Qwen3.6-35B-A3B leads
QwenWebBenchSource 1397Not comparable
CodingQwen3.6-35B-A3B wins
BenchmarkQwen3.5 397BQwen3.6-35B-A3BResult
SWE-bench VerifiedSource 76.2%73.4%Qwen3.5 397B leads
LiveCodeBench v6Source 83.6%Not comparable
SWE-bench ProSource 50.9%49.5%Qwen3.5 397B leads
Terminal-Bench HardSource 40.9%34.8%Qwen3.5 397B leads
AA-SciCodeSource 42.0%35.8%Qwen3.5 397B leads
AA Coding IndexSource 48.2%41.9%Qwen3.5 397B leads
SWE MultilingualSource 67.2%Not comparable
Terminal-Bench 2.0Source 51.5%Not comparable
LiveCodeBenchSource 80.4%Not comparable
NL2RepoSource 29.4%Not comparable
Reasoning
BenchmarkQwen3.5 397BQwen3.6-35B-A3BResult
LongBench v2Source 63.2%Not comparable
AI-NeedleSource 68.7%Not comparable
AA-LCRSource 65.7%63.7%Qwen3.5 397B leads
CritPtSource 1.7%0.3%Qwen3.5 397B leads
KnowledgeQwen3.5 397B wins
BenchmarkQwen3.5 397BQwen3.6-35B-A3BResult
GPQASource 88.4%86%Qwen3.5 397B leads
SuperGPQASource 70.4%64.7%Qwen3.5 397B leads
MMLU-ProSource 87.8%85.2%Qwen3.5 397B leads
MMLU-ReduxSource 94.9%Not comparable
C-EvalSource 93%90%Qwen3.5 397B leads
HLESource 28.7%21.4%Qwen3.5 397B leads
Artificial Analysis Intelligence IndexSource 33.7%31.6%Qwen3.5 397B leads
AA-GPQA DiamondSource 89.3%84.1%Qwen3.5 397B leads
AA-HLESource 27.3%20.2%Qwen3.5 397B leads
AA-Omniscience IndexSource -29.8%-21.4%Qwen3.6-35B-A3B leads
AA-Omniscience AccuracySource 31.4%18.9%Qwen3.5 397B leads
AA-Omniscience Hallucination RateSource 89.1%49.7%Qwen3.6-35B-A3B leads
MathQwen3.5 397B wins
BenchmarkQwen3.5 397BQwen3.6-35B-A3BResult
AIME26Source 93.3%92.7%Qwen3.5 397B leads
HMMT Feb 2025Source 94.8%90.7%Qwen3.5 397B leads
HMMT Nov 2025Source 92.7%89.1%Qwen3.5 397B leads
HMMT Feb 2026Source 87.9%83.6%Qwen3.5 397B leads
MMAnswerBenchSource 80.9%78.9%Qwen3.5 397B leads
Multilingual
BenchmarkQwen3.5 397BQwen3.6-35B-A3BResult
MMLU-ProXSource 84.7%Not comparable
NOVA-63Source 59.1%Not comparable
MultimodalQwen3.5 397B wins
BenchmarkQwen3.5 397BQwen3.6-35B-A3BResult
MMMU-ProSource 79%75.3%Qwen3.5 397B leads
MathVisionSource 88.6%Not comparable
CharXivSource 80.8%78%Qwen3.5 397B leads
VideoMMMUSource 84.7%83.7%Qwen3.5 397B leads
ScreenSpot ProSource 65.6%Not comparable
V*Source 95.8%Not comparable
AA-MMMU-ProSource 77.3%75.0%Qwen3.5 397B leads
MMMUSource 81.7%Not comparable
RealWorldQASource 85.3%Not comparable
OmniDocBench 1.5Source 89.9%Not comparable
SimpleVQASource 58.9%Not comparable
CC-OCRSource 81.9%Not comparable
AI2D_TESTSource 92.7%Not comparable
RefCOCO (avg)Source 92.0%Not comparable
ODINW13Source 50.8%Not comparable
Video-MME (with subtitle)Source 86.6%Not comparable
Video-MME (w/o subtitle)Source 82.5%Not comparable
MLVU (M-Avg)Source 86.2%Not comparable
Inst. Following
BenchmarkQwen3.5 397BQwen3.6-35B-A3BResult
IFEvalSource 92.6%Not comparable
AA-IFBenchSource 78.8%64.4%Qwen3.5 397B leads
Frequently Asked Questions (6)

Which is better, Qwen3.5 397B or Qwen3.6-35B-A3B?

Qwen3.5 397B and Qwen3.6-35B-A3B are tied on the provisional overall score, so the right pick depends on which category matters most for your use case.

Which is better for knowledge tasks, Qwen3.5 397B or Qwen3.6-35B-A3B?

Qwen3.5 397B has the edge for knowledge tasks in this comparison, averaging 56.9 versus 51.8. Inside this category, AA-Omniscience Hallucination Rate is the benchmark that creates the most daylight between them.

Which is better for coding, Qwen3.5 397B or Qwen3.6-35B-A3B?

Qwen3.6-35B-A3B has the edge for coding in this comparison, averaging 73.8 versus 66.5. Inside this category, AA Coding Index is the benchmark that creates the most daylight between them.

Which is better for math, Qwen3.5 397B or Qwen3.6-35B-A3B?

Qwen3.5 397B has the edge for math in this comparison, averaging 90.6 versus 88.2. Inside this category, HMMT Feb 2026 is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Qwen3.5 397B or Qwen3.6-35B-A3B?

Qwen3.5 397B has the edge for agentic tasks in this comparison, averaging 56.5 versus 51.5. Inside this category, GDPval-AA is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, Qwen3.5 397B or Qwen3.6-35B-A3B?

Qwen3.5 397B has the edge for multimodal and grounded tasks in this comparison, averaging 79.6 versus 76.3. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.

Related Comparisons

Last updated: July 15, 2026

The AI models change fast. We track them for you.

A weekly brief for engineers and researchers covering new models, ranking shifts, and pricing changes.

Free. No spam. Unsubscribe anytime.