Skip to main content
BenchLM
Advertise

Model comparison

Claude Sonnet 4.6 vs Qwen3.6 Plus

Data verified

Head-to-head evidence from 25 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

65.07/100
Margin
0.1pts
winning →
65.2/100
2 category wins3 category wins

Public leaderboard positions: Claude Sonnet 4.6 #32 (Supported); Qwen3.6 Plus #30 (Supported). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Claude Sonnet 4.6 and Qwen3.6 Plus share 25 comparable benchmark results. 5 of 8 categories are comparable. 8 results are unique to Claude Sonnet 4.6; 35 to Qwen3.6 Plus.

Updated July 23, 2026
Shared results
25
Claude Sonnet 4.6 only
8
Qwen3.6 Plus only
35
Comparable categories
5 / 8

Pick Qwen3.6 Plus if you want the stronger benchmark profile. Claude Sonnet 4.6 only becomes the better choice if knowledge 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 25 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

Qwen3.6 Plus has the cleaner BenchAlign overall profile here, landing at 65.2 versus 65.07. It is a real lead, but still close enough that category-level strengths matter more than the headline number.

Qwen3.6 Plus's sharpest advantage is in mathematics, where it averages 60.5 against 26.4. The single biggest benchmark swing on the page is SuperGPQA, 95% to 71.6%. Claude Sonnet 4.6 does hit back in knowledge, so the answer changes if that is the part of the workload you care about most.

Qwen3.6 Plus is the reasoning model in the pair, while Claude Sonnet 4.6 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 Plus gives you the larger context window at 1M, compared with 200K for Claude Sonnet 4.6.

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 Claude Sonnet 4.6 and Qwen3.6 Plus
CategoryClaude Sonnet 4.6ΔQwen3.6 Plus
MathClaude Sonnet 4.626.4Margin→ 34.1Qwen3.6 Plus60.5
KnowledgeClaude Sonnet 4.666.0Margin← 8.9Qwen3.6 Plus57.1
AgenticClaude Sonnet 4.665.2Margin← 3.6Qwen3.6 Plus61.6
MultimodalClaude Sonnet 4.677.4Margin→ 2.4Qwen3.6 Plus79.8
CodingClaude Sonnet 4.669.1Margin→ 1.2Qwen3.6 Plus70.3
ReasoningClaude Sonnet 4.6Not measuredMarginNo overlapQwen3.6 Plus62.0
MultilingualClaude Sonnet 4.6Not measuredMarginNo overlapQwen3.6 Plus84.7
Inst. FollowingClaude Sonnet 4.6Not measuredMarginNo overlapQwen3.6 Plus82.3

Decisive benchmark drivers

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

More
A · Claude Sonnet 4.6B · Qwen3.6 Plus

  1. SuperGPQA

    Knowledge
    Source ↗
    A 95%B 71.6%
    Winner: Claude Sonnet 4.6Δ 23.4
    SuperGPQA: Claude Sonnet 4.6 scored 95%; Qwen3.6 Plus scored 71.6%. Claude Sonnet 4.6 wins this benchmark.

  2. HLE

    Knowledge
    Source ↗
    A 49%B 28.8%
    Winner: Claude Sonnet 4.6Δ 20.2
    HLE: Claude Sonnet 4.6 scored 49%; Qwen3.6 Plus scored 28.8%. Claude Sonnet 4.6 wins this benchmark.

  3. MMLU-Pro

    Knowledge
    Source ↗
    A 79.2%B 88.5%
    Winner: Qwen3.6 PlusΔ 9.3
    MMLU-Pro: Claude Sonnet 4.6 scored 79.2%; Qwen3.6 Plus scored 88.5%. Qwen3.6 Plus wins this benchmark.

  4. FrontierMath v2 (Tiers 1-3)

    Math
    Source ↗
    A 32.400%B 26.207%
    Winner: Claude Sonnet 4.6Δ 6.2
    FrontierMath v2 (Tiers 1-3): Claude Sonnet 4.6 scored 32.400%; Qwen3.6 Plus scored 26.207%. Claude Sonnet 4.6 wins this benchmark.

  5. CharXiv

    Multimodal
    Source ↗
    A 77.4%B 81.5%
    Winner: Qwen3.6 PlusΔ 4.1
    CharXiv: Claude Sonnet 4.6 scored 77.4%; Qwen3.6 Plus scored 81.5%. Qwen3.6 Plus wins this benchmark.

Operational comparison

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

MetricClaude Sonnet 4.6Qwen3.6 PlusComparison
Input / output priceUSD per 1M tokensClaude Sonnet 4.6$3 input / $15 outputQwen3.6 PlusNot availableA complete price comparison is not available.
Generation speedtokens per secondClaude Sonnet 4.644 tok/sQwen3.6 PlusNot availableA complete speed comparison is not available.
First-answer latencyseconds to first tokenClaude Sonnet 4.61.48 sQwen3.6 PlusNot availableA complete latency comparison is not available.
Context windowmaximum listed tokensClaude Sonnet 4.6200KQwen3.6 Plus1MQwen3.6 Plus lists the larger context window.

Benchmark Deep Dive

AgenticClaude Sonnet 4.6 wins
BenchmarkClaude Sonnet 4.6Qwen3.6 PlusResult
Terminal-Bench 2.0Source 59.1%61.6%Qwen3.6 Plus leads
OSWorld-VerifiedSource 72.1%—Not comparable
Claw-EvalSource 67.8%58.8%Claude Sonnet 4.6 leads
CyberGymSource 65.2%—Not comparable
τ²-bench resultsSource 79.5%97.7%Qwen3.6 Plus leads
Gert LabsSource 62.92%50.60%Claude Sonnet 4.6 leads
OSWorld 2.0Source 8.3%—Not comparable
JobBenchSource 36.9%—Not comparable
QwenClawBenchSource —57.2%Not comparable
τ³-bench resultsSource —70.7%Not comparable
VITA-BenchSource —44.3%Not comparable
DeepPlanningSource —41.5%Not comparable
ToolathlonSource —39.8%Not comparable
MCP AtlasSource —48.2%Not comparable
MCP-TasksSource —74.1%Not comparable
WideResearchSource —74.3%Not comparable
AA Agentic IndexSource —27.6%Not comparable
GDPval-AASource —31.8%Not comparable
GDPval-AASource —1135Not comparable
ResearchClawBenchSource —18.0%Not comparable
CodingQwen3.6 Plus wins
BenchmarkClaude Sonnet 4.6Qwen3.6 PlusResult
SWE-bench VerifiedSource 79.6%78.8%Claude Sonnet 4.6 leads
SWE-RebenchSource 60.7%—Not comparable
React Native EvalsSource 80.6%—Not comparable
Vibe Code BenchSource 51.48%25.56%Claude Sonnet 4.6 leads
cursorBench31Source 48.8%—Not comparable
AA-SciCodeSource 46.9%40.7%Claude Sonnet 4.6 leads
FrontierCode 1.1 MainSource 24.3%—Not comparable
SWE-bench ProSource —56.6%Not comparable
SWE MultilingualSource —73.8%Not comparable
LiveCodeBench v6Source —87.1%Not comparable
AA Coding IndexSource —54.5%Not comparable
Reasoning
BenchmarkClaude Sonnet 4.6Qwen3.6 PlusResult
AA-LCRSource 57.7%69.7%Qwen3.6 Plus leads
CritPtSource 0.9%2.9%Qwen3.6 Plus leads
AI-NeedleSource —68.3%Not comparable
LongBench v2Source —62%Not comparable
KnowledgeClaude Sonnet 4.6 wins
BenchmarkClaude Sonnet 4.6Qwen3.6 PlusResult
GPQASource 89.9%90.4%Qwen3.6 Plus leads
SuperGPQASource 95%71.6%Claude Sonnet 4.6 leads
MMLU-ProSource 79.2%88.5%Qwen3.6 Plus leads
HLESource 49%28.8%Claude Sonnet 4.6 leads
Artificial Analysis Intelligence IndexSource 35.9%39.6%Qwen3.6 Plus leads
AA-GPQA DiamondSource 79.9%88.2%Qwen3.6 Plus leads
AA-HLESource 13.2%25.7%Qwen3.6 Plus leads
AA-Omniscience IndexSource -2.9%2.7%Qwen3.6 Plus leads
AA-Omniscience AccuracySource 38.0%26.2%Claude Sonnet 4.6 leads
AA-Omniscience Hallucination RateSource 65.9%32.0%Qwen3.6 Plus leads
MMLU-ReduxSource —94.5%Not comparable
C-EvalSource —93.3%Not comparable
MathQwen3.6 Plus wins
BenchmarkClaude Sonnet 4.6Qwen3.6 PlusResult
FrontierMath v2 (Tiers 1-3)Source 32.400%26.207%Claude Sonnet 4.6 leads
FrontierMath v2 (Tier 4)Source 8.300%8.333%Qwen3.6 Plus leads
AIME26Source —95.3%Not comparable
HMMT Feb 2025Source —96.7%Not comparable
HMMT Nov 2025Source —94.6%Not comparable
HMMT Feb 2026Source —87.8%Not comparable
MMAnswerBenchSource —83.8%Not comparable
Multilingual
BenchmarkClaude Sonnet 4.6Qwen3.6 PlusResult
MMLU-ProXSource —84.7%Not comparable
NOVA-63Source —57.9%Not comparable
MultimodalQwen3.6 Plus wins
BenchmarkClaude Sonnet 4.6Qwen3.6 PlusResult
CharXivSource 77.4%81.5%Qwen3.6 Plus leads
AA-MMMU-ProSource 70.6%78.0%Qwen3.6 Plus leads
Design Arena WebsiteSource 13141249Claude Sonnet 4.6 leads
MMMUSource —86.0%Not comparable
MMMU-ProSource —78.8%Not comparable
MathVisionSource —88.0%Not comparable
VideoMMMUSource —84.0%Not comparable
ScreenSpot ProSource —68.2%Not comparable
V*Source —96.9%Not comparable
Inst. Following
BenchmarkClaude Sonnet 4.6Qwen3.6 PlusResult
AA-IFBenchSource 41.2%75.2%Qwen3.6 Plus leads
IFEvalSource —94.3%Not comparable
IFBenchSource —75.8%Not comparable
Frequently Asked Questions (6)

Which is better, Claude Sonnet 4.6 or Qwen3.6 Plus?

Qwen3.6 Plus is ahead on BenchLM's BenchAlign leaderboard, 65.2 to 65.07. The biggest single separator in this matchup is SuperGPQA, where the scores are 95% and 71.6%.

Which is better for knowledge tasks, Claude Sonnet 4.6 or Qwen3.6 Plus?

Claude Sonnet 4.6 has the edge for knowledge tasks in this comparison, averaging 66 versus 57.1. Inside this category, AA-Omniscience Hallucination Rate is the benchmark that creates the most daylight between them.

Which is better for coding, Claude Sonnet 4.6 or Qwen3.6 Plus?

Qwen3.6 Plus has the edge for coding in this comparison, averaging 70.3 versus 69.1. Inside this category, Vibe Code Bench is the benchmark that creates the most daylight between them.

Which is better for math, Claude Sonnet 4.6 or Qwen3.6 Plus?

Qwen3.6 Plus has the edge for math in this comparison, averaging 60.5 versus 26.4. Inside this category, FrontierMath v2 (Tiers 1-3) is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Claude Sonnet 4.6 or Qwen3.6 Plus?

Claude Sonnet 4.6 has the edge for agentic tasks in this comparison, averaging 65.2 versus 61.6. Inside this category, τ²-bench results is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, Claude Sonnet 4.6 or Qwen3.6 Plus?

Qwen3.6 Plus has the edge for multimodal and grounded tasks in this comparison, averaging 79.8 versus 77.4. Inside this category, Design Arena Website is the benchmark that creates the most daylight between them.

Related Comparisons

Explore More

Anthropic Compare Pricing Methodology Find Your Best LLM Overall Rankings
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.