Skip to main content

Model comparison

Qwen3.5-122B-A10B vs Qwen3.5 397B

Data verified

Head-to-head evidence from 28 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

60.56/100
Margin
3.6pts
← winning
57.01/100
3 category wins4 category wins

Public leaderboard positions: Qwen3.5-122B-A10B #47 (Supported); Qwen3.5 397B #71 (Estimated). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Qwen3.5-122B-A10B and Qwen3.5 397B share 28 comparable benchmark results. 7 of 8 categories are comparable. 3 results are unique to Qwen3.5-122B-A10B; 27 to Qwen3.5 397B.

Updated July 23, 2026
Shared results
28
Qwen3.5-122B-A10B only
3
Qwen3.5 397B only
27
Comparable categories
7 / 8

Pick Qwen3.5-122B-A10B if you want the stronger benchmark profile. Qwen3.5 397B only becomes the better choice if reasoning is the priority or you would rather avoid the extra latency and token burn of a reasoning model.

Confidence note. This is a partial-evidence comparison with 28 shared benchmark results across 7 evidence categories; 7 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

Qwen3.5-122B-A10B is clearly ahead on the BenchAlign aggregate, 60.56 to 57.01. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

Qwen3.5-122B-A10B's sharpest advantage is in knowledge, where it averages 83.6 against 56.6. The single biggest benchmark swing on the page is SWE-bench Verified, 72% to 76.2%. Qwen3.5 397B does hit back in reasoning, so the answer changes if that is the part of the workload you care about most.

Qwen3.5 397B is also the more expensive model on tokens at $0.60 input / $3.60 output per 1M tokens, versus $0.00 input / $0.00 output per 1M tokens for Qwen3.5-122B-A10B. That is roughly Infinityx on output cost alone. Qwen3.5-122B-A10B 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. Qwen3.5-122B-A10B 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 Qwen3.5-122B-A10B and Qwen3.5 397B
CategoryQwen3.5-122B-A10BΔQwen3.5 397B
KnowledgeQwen3.5-122B-A10B83.6Margin 27.0Qwen3.5 397B56.6
CodingQwen3.5-122B-A10B72.0Margin 5.5Qwen3.5 397B66.5
ReasoningQwen3.5-122B-A10B60.2Margin 3.0Qwen3.5 397B63.2
MultilingualQwen3.5-122B-A10B82.2Margin 2.5Qwen3.5 397B84.7
MultimodalQwen3.5-122B-A10B77.2Margin 2.4Qwen3.5 397B79.6
Inst. FollowingQwen3.5-122B-A10B93.4Margin 0.8Qwen3.5 397B92.6
AgenticQwen3.5-122B-A10B56.4Margin 0.1Qwen3.5 397B56.5
MathQwen3.5-122B-A10BNot measuredMarginNo overlapQwen3.5 397B90.6

Decisive benchmark drivers

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

More
A · Qwen3.5-122B-A10BB · Qwen3.5 397B
  1. SWE-bench Verified

    Coding
    Source ↗
    A 72%B 76.2%
    Winner: Qwen3.5 397BΔ 4.2
    SWE-bench Verified: Qwen3.5-122B-A10B scored 72%; Qwen3.5 397B scored 76.2%. Qwen3.5 397B wins this benchmark.
  2. CharXiv

    Multimodal
    Source ↗
    A 77.2%B 80.8%
    Winner: Qwen3.5 397BΔ 3.6
    CharXiv: Qwen3.5-122B-A10B scored 77.2%; Qwen3.5 397B scored 80.8%. Qwen3.5 397B wins this benchmark.
  3. SuperGPQA

    Knowledge
    Source ↗
    A 67.1%B 70.4%
    Winner: Qwen3.5 397BΔ 3.3
    SuperGPQA: Qwen3.5-122B-A10B scored 67.1%; Qwen3.5 397B scored 70.4%. Qwen3.5 397B wins this benchmark.
  4. Terminal-Bench 2.0

    Agentic
    Source ↗
    A 49.4%B 52.5%
    Winner: Qwen3.5 397BΔ 3.1
    Terminal-Bench 2.0: Qwen3.5-122B-A10B scored 49.4%; Qwen3.5 397B scored 52.5%. Qwen3.5 397B wins this benchmark.
  5. LongBench v2

    Reasoning
    Source ↗
    A 60.2%B 63.2%
    Winner: Qwen3.5 397BΔ 3
    LongBench v2: Qwen3.5-122B-A10B scored 60.2%; Qwen3.5 397B scored 63.2%. Qwen3.5 397B wins this benchmark.

Operational comparison

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

MetricQwen3.5-122B-A10BQwen3.5 397BComparison
Input / output priceUSD per 1M tokensQwen3.5-122B-A10B$0 input / $0 outputQwen3.5 397B$0.6 input / $3.6 outputQwen3.5-122B-A10B has the lower combined listed price.
Generation speedtokens per secondQwen3.5-122B-A10BNot availableQwen3.5 397B96 tok/sA complete speed comparison is not available.
First-answer latencyseconds to first tokenQwen3.5-122B-A10BNot availableQwen3.5 397B2.44 sA complete latency comparison is not available.
Context windowmaximum listed tokensQwen3.5-122B-A10B262KQwen3.5 397B128KQwen3.5-122B-A10B lists the larger context window.

Benchmark Deep Dive

AgenticQwen3.5 397B wins
BenchmarkQwen3.5-122B-A10BQwen3.5 397BResult
Terminal-Bench 2.0Source 49.4%52.5%Qwen3.5 397B leads
BrowseCompSource 63.8%62%Qwen3.5-122B-A10B leads
OSWorld-VerifiedSource 58%Not comparable
τ²-bench resultsSource 93.6%95.6%Qwen3.5 397B leads
AA Agentic IndexSource 20.7%19.9%Qwen3.5-122B-A10B leads
GDPval-AASource 23.9%23.1%Qwen3.5-122B-A10B leads
GDPval-AASource 978962Qwen3.5-122B-A10B leads
Claw-EvalSource 56.8%Not comparable
QwenClawBenchSource 51.8%Not comparable
τ³-bench resultsSource 68.4%Not comparable
VITA-BenchSource 43.7%Not comparable
DeepPlanningSource 37.6%Not comparable
ToolathlonSource 36.3%Not comparable
MCP AtlasSource 46.1%Not comparable
MCP-TasksSource 74.2%Not comparable
WideResearchSource 74.0%Not comparable
Gert LabsSource 46.76%Not comparable
ResearchClawBenchSource 14.2%Not comparable
APEX-Agents-AASource 15.3%Not comparable
CodingQwen3.5-122B-A10B wins
BenchmarkQwen3.5-122B-A10BQwen3.5 397BResult
SWE-bench VerifiedSource 72%76.2%Qwen3.5 397B leads
AA Coding IndexSource 45.7%48.2%Qwen3.5 397B leads
AA-SciCodeSource 42.0%42.0%Tie
LiveCodeBench v6Source 83.6%Not comparable
SWE-bench ProSource 50.9%Not comparable
ReasoningQwen3.5 397B wins
BenchmarkQwen3.5-122B-A10BQwen3.5 397BResult
LongBench v2Source 60.2%63.2%Qwen3.5 397B leads
AA-LCRSource 66.7%65.7%Qwen3.5-122B-A10B leads
CritPtSource 0.6%1.7%Qwen3.5 397B leads
AI-NeedleSource 68.7%Not comparable
KnowledgeQwen3.5-122B-A10B wins
BenchmarkQwen3.5-122B-A10BQwen3.5 397BResult
MMLU-ProSource 86.7%87.8%Qwen3.5 397B leads
SuperGPQASource 67.1%70.4%Qwen3.5 397B leads
GPQASource 86.6%88.4%Qwen3.5 397B leads
Artificial Analysis Intelligence IndexSource 32.3%33.7%Qwen3.5 397B leads
AA-GPQA DiamondSource 85.7%89.3%Qwen3.5 397B leads
AA-HLESource 23.4%27.3%Qwen3.5 397B leads
AA-Omniscience IndexSource -39.6%-29.8%Qwen3.5 397B leads
AA-Omniscience AccuracySource 24.7%31.4%Qwen3.5 397B leads
AA-Omniscience Hallucination RateSource 85.5%89.1%Qwen3.5-122B-A10B leads
MMLU-ReduxSource 94.9%Not comparable
C-EvalSource 93%Not comparable
HLESource 28.7%Not comparable
Math
BenchmarkQwen3.5-122B-A10BQwen3.5 397BResult
AIME26Source 93.3%Not comparable
HMMT Feb 2025Source 94.8%Not comparable
HMMT Nov 2025Source 92.7%Not comparable
HMMT Feb 2026Source 87.9%Not comparable
MMAnswerBenchSource 80.9%Not comparable
MultilingualQwen3.5 397B wins
BenchmarkQwen3.5-122B-A10BQwen3.5 397BResult
MMLU-ProXSource 82.2%84.7%Qwen3.5 397B leads
NOVA-63Source 59.1%Not comparable
MultimodalQwen3.5 397B wins
BenchmarkQwen3.5-122B-A10BQwen3.5 397BResult
MMMUSource 83.9%Not comparable
MMVUSource 74.7%Not comparable
MathVisionSource 86.2%88.6%Qwen3.5 397B leads
CharXivSource 77.2%80.8%Qwen3.5 397B leads
V*Source 93.2%95.8%Qwen3.5 397B leads
AA-MMMU-ProSource 75.0%77.3%Qwen3.5 397B leads
MMMU-ProSource 79%Not comparable
VideoMMMUSource 84.7%Not comparable
ScreenSpot ProSource 65.6%Not comparable
Inst. FollowingQwen3.5-122B-A10B wins
BenchmarkQwen3.5-122B-A10BQwen3.5 397BResult
IFEvalSource 93.4%92.6%Qwen3.5-122B-A10B leads
AA-IFBenchSource 75.7%78.8%Qwen3.5 397B leads
Frequently Asked Questions (8)

Which is better, Qwen3.5-122B-A10B or Qwen3.5 397B?

Qwen3.5-122B-A10B is ahead on BenchLM's BenchAlign leaderboard, 60.56 to 57.01. The biggest single separator in this matchup is SWE-bench Verified, where the scores are 72% and 76.2%.

Which is better for knowledge tasks, Qwen3.5-122B-A10B or Qwen3.5 397B?

Qwen3.5-122B-A10B has the edge for knowledge tasks in this comparison, averaging 83.6 versus 56.6. Inside this category, AA-Omniscience Index is the benchmark that creates the most daylight between them.

Which is better for coding, Qwen3.5-122B-A10B or Qwen3.5 397B?

Qwen3.5-122B-A10B has the edge for coding in this comparison, averaging 72 versus 66.5. Inside this category, SWE-bench Verified is the benchmark that creates the most daylight between them.

Which is better for reasoning, Qwen3.5-122B-A10B or Qwen3.5 397B?

Qwen3.5 397B has the edge for reasoning in this comparison, averaging 63.2 versus 60.2. Inside this category, LongBench v2 is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Qwen3.5-122B-A10B or Qwen3.5 397B?

Qwen3.5 397B has the edge for agentic tasks in this comparison, averaging 56.5 versus 56.4. Inside this category, GDPval-AA is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, Qwen3.5-122B-A10B or Qwen3.5 397B?

Qwen3.5 397B has the edge for multimodal and grounded tasks in this comparison, averaging 79.6 versus 77.2. Inside this category, CharXiv is the benchmark that creates the most daylight between them.

Which is better for instruction following, Qwen3.5-122B-A10B or Qwen3.5 397B?

Qwen3.5-122B-A10B has the edge for instruction following in this comparison, averaging 93.4 versus 92.6. Inside this category, AA-IFBench is the benchmark that creates the most daylight between them.

Which is better for multilingual tasks, Qwen3.5-122B-A10B or Qwen3.5 397B?

Qwen3.5 397B has the edge for multilingual tasks in this comparison, averaging 84.7 versus 82.2. Inside this category, MMLU-ProX is the benchmark that creates the most daylight between them.

Related Comparisons

Last updated: July 23, 2026

Choose a model with this week’s evidence

Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.

One email each week. Unsubscribe anytime.