Skip to main content
Radar

Five or fewer confirmed AI changes, with original sources, on mornings when something changed.A free source-linked morning brief.

Start free brief
BenchLM
Model A
Qwen3.5 397B

Alibaba

56.5/100

Estimated · Public rank #80

90% interval 45.0–68.0

Qwen3.5 397B vs Sakana Fugu-Ultra

Updated August 13, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.

Model B
Sakana Fugu-Ultra

Sakana AI

—

Evidence status unavailable

90% interval unavailable

Decision reading

At least one model is not scored in the current public ranking lane, so the page does not name an overall quality winner.

5 results are shared. Category rows based on different benchmark sets are marked directional and do not name a winner.

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

    Sakana Fugu-Ultra

    Sakana Fugu-Ultra 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

    The category averages use different weighted benchmark sets, so they are directional rather than like-for-like.

    Confidence: limited

  • Agentic work

    Tool use, computer use, and multi-step task completion

    Not enough matched evidence

    The category averages use different weighted benchmark sets, so they are 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. Sakana Fugu-Ultra 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.

Shared results
5
Qwen3.5 397B only
33
Sakana Fugu-Ultra only
6
Like-for-like categories
0 / 8

4 categories use different evidence sets. Those rows remain visible for coverage context but do not name a winner.

Category results, on a stated basis

Each row states whether both averages use the same weighted benchmark set. Directional and not-comparable rows remain visible, but they never receive a winner in this template.

Agentic

Directional only
Qwen3.5 397B
56.5
Sakana Fugu-Ultra
82.1
Weighted basis
2 vs 1 rows
Reading
Directional only

Coding

Directional only
Qwen3.5 397B
66.5
Sakana Fugu-Ultra
64.5
Weighted basis
2 vs 2 rows
Reading
Directional only

Knowledge

Directional only
Qwen3.5 397B
56.6
Sakana Fugu-Ultra
95.5
Weighted basis
4 vs 1 rows
Reading
Directional only

Multimodal

Directional only
Qwen3.5 397B
79.6
Sakana Fugu-Ultra
86.6
Weighted basis
2 vs 1 rows
Reading
Directional only

Reasoning

Not comparable
Qwen3.5 397B
63.2
Sakana Fugu-Ultra
93.6
Weighted basis
1 vs 1 rows
Reading
Not comparable

Math

Not comparable
Qwen3.5 397B
90.6
Sakana Fugu-Ultra
Not measured
Weighted basis
2 vs 0 rows
Reading
Not comparable

Multilingual

Not comparable
Qwen3.5 397B
84.7
Sakana Fugu-Ultra
Not measured
Weighted basis
1 vs 0 rows
Reading
Not comparable

Instruction following

Not comparable
Qwen3.5 397B
92.6
Sakana Fugu-Ultra
Not measured
Weighted basis
1 vs 0 rows
Reading
Not comparable

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.

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
Sakana Fugu-Ultra
API rate not published
Fits in one request

Sakana Fugu-Ultra has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

Qwen3.5 397B
$0.0408
Fits in one request
Sakana Fugu-Ultra
API rate not published
Fits in one request

Sakana Fugu-Ultra 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
Sakana Fugu-Ultra
API rate not published
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. Sakana Fugu-Ultra 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.

Context window

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

Qwen3.5 397B

128K

Sakana Fugu-Ultra

1M

API model ID

Qwen3.5 397B

Not sourced

Sakana Fugu-Ultra

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

Sakana Fugu-Ultra

No comparable hosted API rate

Documented inputs

Qwen3.5 397B

Not sourced

Sakana Fugu-Ultra

Not sourced

Documented outputs

Qwen3.5 397B

Not sourced

Sakana Fugu-Ultra

Not sourced

Provider availability

Qwen3.5 397B

Not sourced

Sakana Fugu-Ultra

Not sourced

Reasoning profile

Qwen3.5 397B

Non-Reasoning

Sakana Fugu-Ultra

Reasoning

Weight access

Qwen3.5 397B

Open Weight

Sakana Fugu-Ultra

Proprietary

License

Qwen3.5 397B

Open Weight

Sakana Fugu-Ultra

Proprietary

Release date

Qwen3.5 397B

2026-02-16

Sakana Fugu-Ultra

2026-06-22

If you already use one of these models
Deployment change
The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
Quality signal
At least one model is not scored in the current public ranking lane, so the page does not name an overall quality winner.
Workload cost
A complete comparable API-rate estimate is not available for both models.
Context tradeoff
Sakana Fugu-Ultra has the larger documented window (1M).

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

Benchmark evidence

The full public result ledger is available for audit without forcing a wide desktop table onto a phone.

Browse raw public benchmark evidence44 rows

Agentic

  • Terminal-Bench 2.0

    Qwen3.5 397B52.5%
    Source
    Sakana Fugu-Ultra82.1%
    Source

    Sakana Fugu-Ultra leads this result

  • BrowseComp

    Qwen3.5 397B62%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • Claw-Eval

    Qwen3.5 397B56.8%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • QwenClawBench

    Qwen3.5 397B51.8%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • τ³-bench results

    Qwen3.5 397B68.4%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • VITA-Bench

    Qwen3.5 397B43.7%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • DeepPlanning

    Qwen3.5 397B37.6%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • Toolathlon

    Qwen3.5 397B36.3%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • MCP Atlas

    Qwen3.5 397B46.1%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • MCP-Tasks

    Qwen3.5 397B74.2%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • WideResearch

    Qwen3.5 397B74.0%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • Gert Labs

    Qwen3.5 397B46.76%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • ResearchClawBench

    Qwen3.5 397B14.2%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

Coding

  • SWE-bench Verified

    Qwen3.5 397B76.2%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • LiveCodeBench v6

    Qwen3.5 397B83.6%
    Source
    Sakana Fugu-Ultra93.2%
    Source

    Sakana Fugu-Ultra leads this result

  • SWE-bench Pro

    Qwen3.5 397B50.9%
    Source
    Sakana Fugu-Ultra73.7%
    Source

    Sakana Fugu-Ultra leads this result

  • Terminal-Bench 2.0

    Qwen3.5 397B—
    Sakana Fugu-Ultra82.1%
    Source

    Not directly comparable

  • LiveCodeBench Pro

    Qwen3.5 397B—
    Sakana Fugu-Ultra90.8%
    Source

    Not directly comparable

  • SciCode

    Qwen3.5 397B—
    Sakana Fugu-Ultra58.7%
    Source

    Not directly comparable

Reasoning

  • LongBench v2

    Qwen3.5 397B63.2%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • AI-Needle

    Qwen3.5 397B68.7%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • MRCRv2

    Qwen3.5 397B—
    Sakana Fugu-Ultra93.6%
    Source

    Not directly comparable

Knowledge

  • GPQA

    Qwen3.5 397B88.4%
    Source
    Sakana Fugu-Ultra95.5%
    Source

    Sakana Fugu-Ultra leads this result

  • SuperGPQA

    Qwen3.5 397B70.4%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • MMLU-Pro

    Qwen3.5 397B87.8%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • MMLU-Redux

    Qwen3.5 397B94.9%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • C-Eval

    Qwen3.5 397B93%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • HLE

    Qwen3.5 397B28.7%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • GPQA-D

    Qwen3.5 397B—
    Sakana Fugu-Ultra95.5%
    Source

    Not directly comparable

  • HLE w/o tools

    Qwen3.5 397B—
    Sakana Fugu-Ultra50%
    Source

    Not directly comparable

Math

  • AIME26

    Qwen3.5 397B93.3%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • HMMT Feb 2025

    Qwen3.5 397B94.8%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • HMMT Nov 2025

    Qwen3.5 397B92.7%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • HMMT Feb 2026

    Qwen3.5 397B87.9%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • MMAnswerBench

    Qwen3.5 397B80.9%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

Multilingual

  • MMLU-ProX

    Qwen3.5 397B84.7%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • NOVA-63

    Qwen3.5 397B59.1%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

Multimodal

  • MMMU-Pro

    Qwen3.5 397B79%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • MathVision

    Qwen3.5 397B88.6%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • CharXiv

    Qwen3.5 397B80.8%
    Source
    Sakana Fugu-Ultra86.6%
    Source

    Sakana Fugu-Ultra leads this result

  • VideoMMMU

    Qwen3.5 397B84.7%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • ScreenSpot Pro

    Qwen3.5 397B65.6%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

  • V*

    Qwen3.5 397B95.8%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

Instruction following

  • IFEval

    Qwen3.5 397B92.6%
    Source
    Sakana Fugu-Ultra—

    Not directly comparable

Frequently asked questions

Which is better, Qwen3.5 397B or Sakana Fugu-Ultra?

At least one model is not scored in the current public ranking lane, so the page does not name an overall quality winner. The page therefore keeps the decision tied to the specific documented workload.

Which is better for coding, Qwen3.5 397B or Sakana Fugu-Ultra?

The current coding averages use different weighted benchmark sets, so BenchLM does not name a winner from them. 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 Sakana Fugu-Ultra?

The current agentic tasks averages use different weighted benchmark sets, so BenchLM does not name a winner from them. Read the shared benchmark rows directly and test the models on the same task set.

Which costs less, Qwen3.5 397B or Sakana Fugu-Ultra?

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 Sakana Fugu-Ultra?

Sakana Fugu-Ultra has the larger documented context window: 1M, compared with 128K.

Related comparisons

Compare API pricing Read the methodology Open the model selector

Last updated August 13, 2026

Watch Qwen3.5 397B vs Sakana Fugu-Ultra

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

Read a sample issue

Join 2,000+ readers.