Skip to main content

Model comparison

Qwen3.7 Max vs Sakana Fugu

Data verified

Head-to-head evidence from 7 shared benchmark results across 4 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

72.84/100
No comparison
Sakana AI
N/A
2 category wins2 category wins

Public leaderboard positions: Qwen3.7 Max #10 (Supported); Sakana Fugu unranked (Not scored). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Qwen3.7 Max and Sakana Fugu share 7 comparable benchmark results. 4 of 8 categories are comparable. 51 results are unique to Qwen3.7 Max; 4 to Sakana Fugu.

Updated July 23, 2026
Shared results
7
Qwen3.7 Max only
51
Sakana Fugu only
4
Comparable categories
4 / 8

Treat this as a split decision. Qwen3.7 Max makes more sense if coding is the priority; Sakana Fugu is the better fit if knowledge is the priority.

Confidence note. This is a partial-evidence comparison with 7 shared benchmark results across 4 evidence categories; 4 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

Qwen3.7 Max and Sakana Fugu finish on the same BenchAlign overall score, so this is less about a single winner and more about where the edge shows up. The BenchAlign headline says tie; the benchmark table is where the real choice happens.

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 Qwen3.7 Max and Sakana Fugu
CategoryQwen3.7 MaxΔSakana Fugu
KnowledgeQwen3.7 Max64.2Margin 31.3Sakana Fugu95.5
CodingQwen3.7 Max77.9Margin 18.2Sakana Fugu59.7
AgenticQwen3.7 Max69.7Margin 10.5Sakana Fugu80.2
ReasoningQwen3.7 Max90.4Margin 3.8Sakana Fugu86.6
MathQwen3.7 Max97.1MarginNo overlapSakana FuguNot measured
MultilingualQwen3.7 Max87.0MarginNo overlapSakana FuguNot measured
MultimodalQwen3.7 MaxNot measuredMarginNo overlapSakana Fugu85.1
Inst. FollowingQwen3.7 Max84.4MarginNo overlapSakana FuguNot measured

Decisive benchmark drivers

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

More
A · Qwen3.7 MaxB · Sakana Fugu
  1. Terminal-Bench 2.0

    Agentic
    Source ↗
    A 69.7%B 80.2%
    Winner: Sakana FuguΔ 10.5
    Terminal-Bench 2.0: Qwen3.7 Max scored 69.7%; Sakana Fugu scored 80.2%. Sakana Fugu wins this benchmark.
  2. SciCode

    Coding
    Source ↗
    A 53.5%B 60.1%
    Winner: Sakana FuguΔ 6.6
    SciCode: Qwen3.7 Max scored 53.5%; Sakana Fugu scored 60.1%. Sakana Fugu wins this benchmark.
  3. MRCRv2

    Reasoning
    Source ↗
    A 90.4%B 86.6%
    Winner: Qwen3.7 MaxΔ 3.8
    MRCRv2: Qwen3.7 Max scored 90.4%; Sakana Fugu scored 86.6%. Qwen3.7 Max wins this benchmark.
  4. GPQA

    Knowledge
    Source ↗
    A 92.4%B 95.5%
    Winner: Sakana FuguΔ 3.1
    GPQA: Qwen3.7 Max scored 92.4%; Sakana Fugu scored 95.5%. Sakana Fugu wins this benchmark.
  5. SWE-bench Pro

    Coding
    Source ↗
    A 60.6%B 59%
    Winner: Qwen3.7 MaxΔ 1.6
    SWE-bench Pro: Qwen3.7 Max scored 60.6%; Sakana Fugu scored 59%. Qwen3.7 Max wins this benchmark.

Operational comparison

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

MetricQwen3.7 MaxSakana FuguComparison
Input / output priceUSD per 1M tokensQwen3.7 MaxNot availableSakana FuguNot availableA complete price comparison is not available.
Generation speedtokens per secondQwen3.7 MaxNot availableSakana FuguNot availableA complete speed comparison is not available.
First-answer latencyseconds to first tokenQwen3.7 MaxNot availableSakana FuguNot availableA complete latency comparison is not available.
Context windowmaximum listed tokensQwen3.7 Max1MSakana Fugu1MListed context windows are equal.

Benchmark Deep Dive

AgenticSakana Fugu wins
BenchmarkQwen3.7 MaxSakana FuguResult
Terminal-Bench 2.0Source 69.7%80.2%Sakana Fugu leads
QwenClawBenchSource 64.3%Not comparable
QwenWebBenchSource 1568Not comparable
Claw-EvalSource 65.2%Not comparable
BFCL v4Source 75.0%Not comparable
MCP AtlasSource 76.4%Not comparable
VITA-BenchSource 47.9%Not comparable
HLE w/ toolsSource 53.5%Not comparable
AA Agentic IndexSource 30.6%Not comparable
τ²-bench resultsSource 94.7%Not comparable
GDPval-AASource 38.7%Not comparable
GDPval-AASource 1273Not comparable
Gert LabsSource 64.27%Not comparable
ResearchClawBenchSource 18.7%Not comparable
AA BriefcaseSource 908Not comparable
AA AutomationBenchSource 25.6%Not comparable
AA EnterpriseOps-GymSource 45.0%Not comparable
AA ITBenchSource 42.5%Not comparable
terminalBenchHardSource 50.8%Not comparable
aaTerminalBench21Source 74.5%Not comparable
AA Harvey LABSource 83.4%Not comparable
CodingQwen3.7 Max wins
BenchmarkQwen3.7 MaxSakana FuguResult
SWE-bench VerifiedSource 80.4%Not comparable
SWE-bench ProSource 60.6%59%Qwen3.7 Max leads
SWE MultilingualSource 78.3%Not comparable
NL2RepoSource 47.2%Not comparable
SciCodeSource 53.5%60.1%Sakana Fugu leads
LiveCodeBenchSource 91.6%Not comparable
Terminal-Bench 2.0Source 69.7%80.2%Sakana Fugu leads
AA Coding IndexSource 66.0%Not comparable
AA-SciCodeSource 48.8%Not comparable
LiveCodeBench v6Source 92.9%Not comparable
LiveCodeBench ProSource 87.8%Not comparable
ReasoningQwen3.7 Max wins
BenchmarkQwen3.7 MaxSakana FuguResult
MRCRv2Source 90.4%86.6%Qwen3.7 Max leads
CritPtSource 13.4%Not comparable
AA-LCRSource 69.0%Not comparable
KnowledgeSakana Fugu wins
BenchmarkQwen3.7 MaxSakana FuguResult
GPQASource 92.4%95.5%Sakana Fugu leads
GPQA-DSource 92.4%95.5%Sakana Fugu leads
HLESource 41.4%Not comparable
MMLU-ProSource 89.6%Not comparable
MMLU-ReduxSource 95%Not comparable
SuperGPQASource 73.6%Not comparable
MMMLUSource 90.3%Not comparable
Artificial Analysis Intelligence IndexSource 46.0%Not comparable
AA-GPQA DiamondSource 92.3%Not comparable
AA-HLESource 38.1%Not comparable
AA-Omniscience IndexSource 14.1%Not comparable
AA-Omniscience AccuracySource 30.1%Not comparable
AA-Omniscience Hallucination RateSource 22.9%Not comparable
HLE w/o toolsSource 47.2%Not comparable
Math
BenchmarkQwen3.7 MaxSakana FuguResult
HMMT Feb 2026Source 97.1%Not comparable
IMOAnswerBenchSource 90.0%Not comparable
ApexSource 44.5%Not comparable
Multilingual
BenchmarkQwen3.7 MaxSakana FuguResult
MMLU-ProXSource 87%Not comparable
NOVA-63Source 59.0%Not comparable
INCLUDESource 86.2%Not comparable
MAXIFESource 89.2%Not comparable
PolyMathSource 86.5%Not comparable
Multimodal
BenchmarkQwen3.7 MaxSakana FuguResult
Design Arena WebsiteSource 1293Not comparable
CharXivSource 85.1%Not comparable
Inst. Following
BenchmarkQwen3.7 MaxSakana FuguResult
IFEvalSource 94.3%Not comparable
IFBenchSource 79.1%Not comparable
AA-IFBenchSource 80.5%Not comparable
Frequently Asked Questions (5)

Which is better, Qwen3.7 Max or Sakana Fugu?

Qwen3.7 Max and Sakana Fugu are tied on the BenchAlign overall score, so the right pick depends on which category matters most for your use case.

Which is better for knowledge tasks, Qwen3.7 Max or Sakana Fugu?

Sakana Fugu has the edge for knowledge tasks in this comparison, averaging 95.5 versus 64.2. Inside this category, GPQA is the benchmark that creates the most daylight between them.

Which is better for coding, Qwen3.7 Max or Sakana Fugu?

Qwen3.7 Max has the edge for coding in this comparison, averaging 77.9 versus 59.7. Inside this category, Terminal-Bench 2.0 is the benchmark that creates the most daylight between them.

Which is better for reasoning, Qwen3.7 Max or Sakana Fugu?

Qwen3.7 Max has the edge for reasoning in this comparison, averaging 90.4 versus 86.6. Inside this category, MRCRv2 is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Qwen3.7 Max or Sakana Fugu?

Sakana Fugu has the edge for agentic tasks in this comparison, averaging 80.2 versus 69.7. Inside this category, Terminal-Bench 2.0 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.