Skip to main content

Model comparison

Kimi K2.5 vs Sakana Fugu-Ultra

Data verified

Head-to-head evidence from 6 shared benchmark results across 3 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

Moonshot AI
59.66/100
No comparison
N/A
0 category wins5 category wins

Public leaderboard positions: Kimi K2.5 #54 (Supported); Sakana Fugu-Ultra unranked (Not scored). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Kimi K2.5 and Sakana Fugu-Ultra share 6 comparable benchmark results. 5 of 8 categories are comparable. 57 results are unique to Kimi K2.5; 5 to Sakana Fugu-Ultra.

Updated July 23, 2026
Shared results
6
Kimi K2.5 only
57
Sakana Fugu-Ultra only
5
Comparable categories
5 / 8

Treat this as a split decision. Kimi K2.5 makes more sense if you would rather avoid the extra latency and token burn of a reasoning model; Sakana Fugu-Ultra is the better fit if knowledge is the priority or you need the larger 1M context window.

Confidence note. This is a partial-evidence comparison with 6 shared benchmark results across 3 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

Kimi K2.5 and Sakana Fugu-Ultra 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.

Sakana Fugu-Ultra is the reasoning model in the pair, while Kimi K2.5 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. Sakana Fugu-Ultra gives you the larger context window at 1M, compared with 256K for Kimi K2.5.

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 Kimi K2.5 and Sakana Fugu-Ultra
CategoryKimi K2.5ΔSakana Fugu-Ultra
KnowledgeKimi K2.556.9Margin 38.6Sakana Fugu-Ultra95.5
ReasoningKimi K2.561.0Margin 32.6Sakana Fugu-Ultra93.6
AgenticKimi K2.555.0Margin 27.1Sakana Fugu-Ultra82.1
MultimodalKimi K2.578.5Margin 8.1Sakana Fugu-Ultra86.6
CodingKimi K2.559.4Margin 5.1Sakana Fugu-Ultra64.5
MathKimi K2.560.6MarginNo overlapSakana Fugu-UltraNot measured
MultilingualKimi K2.582.3MarginNo overlapSakana Fugu-UltraNot measured
Inst. FollowingKimi K2.593.9MarginNo overlapSakana Fugu-UltraNot measured

Decisive benchmark drivers

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

More
A · Kimi K2.5B · Sakana Fugu-Ultra
  1. Terminal-Bench 2.0

    Agentic
    Source ↗
    A 50.8%B 82.1%
    Winner: Sakana Fugu-UltraΔ 31.3
    Terminal-Bench 2.0: Kimi K2.5 scored 50.8%; Sakana Fugu-Ultra scored 82.1%. Sakana Fugu-Ultra wins this benchmark.
  2. SWE-bench Pro

    Coding
    Source ↗
    A 50.7%B 73.7%
    Winner: Sakana Fugu-UltraΔ 23
    SWE-bench Pro: Kimi K2.5 scored 50.7%; Sakana Fugu-Ultra scored 73.7%. Sakana Fugu-Ultra wins this benchmark.
  3. SciCode

    Coding
    Source ↗
    A 48.7%B 58.7%
    Winner: Sakana Fugu-UltraΔ 10
    SciCode: Kimi K2.5 scored 48.7%; Sakana Fugu-Ultra scored 58.7%. Sakana Fugu-Ultra wins this benchmark.
  4. GPQA

    Knowledge
    Source ↗
    A 87.6%B 95.5%
    Winner: Sakana Fugu-UltraΔ 7.9
    GPQA: Kimi K2.5 scored 87.6%; Sakana Fugu-Ultra scored 95.5%. Sakana Fugu-Ultra wins this benchmark.

Operational comparison

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

MetricKimi K2.5Sakana Fugu-UltraComparison
Input / output priceUSD per 1M tokensKimi K2.5$0.6 input / $3 outputSakana Fugu-UltraNot availableA complete price comparison is not available.
Generation speedtokens per secondKimi K2.545 tok/sSakana Fugu-UltraNot availableA complete speed comparison is not available.
First-answer latencyseconds to first tokenKimi K2.52.38 sSakana Fugu-UltraNot availableA complete latency comparison is not available.
Context windowmaximum listed tokensKimi K2.5256KSakana Fugu-Ultra1MSakana Fugu-Ultra lists the larger context window.

Benchmark Deep Dive

AgenticSakana Fugu-Ultra wins
BenchmarkKimi K2.5Sakana Fugu-UltraResult
Terminal-Bench 2.0Source 50.8%82.1%Sakana Fugu-Ultra leads
BrowseCompSource 60.6%Not comparable
Claw-EvalSource 52.3%Not comparable
QwenClawBenchSource 54.3%Not comparable
τ³-bench resultsSource 65.7%Not comparable
DeepSearchQASource 77.1%Not comparable
DeepPlanningSource 14.4%Not comparable
ToolathlonSource 27.8%Not comparable
MCP AtlasSource 29.5%Not comparable
MCP-TasksSource 59.1%Not comparable
WideResearchSource 72.7%Not comparable
τ²-bench resultsSource 95.9%Not comparable
APEX-Agents-AASource 11.5%Not comparable
Gert LabsSource 45.88%Not comparable
ResearchClawBenchSource 14.0%Not comparable
JobBenchSource 8.7%Not comparable
AA Agentic IndexSource 21.7%Not comparable
GDPval-AASource 25.4%Not comparable
GDPval-AASource 1009Not comparable
CodingSakana Fugu-Ultra wins
BenchmarkKimi K2.5Sakana Fugu-UltraResult
SWE-bench VerifiedSource 76.8%Not comparable
SWE-bench Verified*Source 70.8%Not comparable
LiveCodeBench v6Source 85.0%93.2%Sakana Fugu-Ultra leads
SWE-bench ProSource 50.7%73.7%Sakana Fugu-Ultra leads
SWE MultilingualSource 73%Not comparable
SWE-RebenchSource 58.5%Not comparable
React Native EvalsSource 77.2%Not comparable
SciCodeSource 48.7%58.7%Sakana Fugu-Ultra leads
AA-SciCodeSource 49.0%Not comparable
AA Coding IndexSource 46.8%Not comparable
Terminal-Bench 2.0Source 82.1%Not comparable
LiveCodeBench ProSource 90.8%Not comparable
ReasoningSakana Fugu-Ultra wins
BenchmarkKimi K2.5Sakana Fugu-UltraResult
LongBench v2Source 61%Not comparable
AA-LCRSource 65.3%Not comparable
CritPtSource 3.1%Not comparable
MRCRv2Source 93.6%Not comparable
KnowledgeSakana Fugu-Ultra wins
BenchmarkKimi K2.5Sakana Fugu-UltraResult
GPQASource 87.6%95.5%Sakana Fugu-Ultra leads
GPQA-DSource 87.6%95.5%Sakana Fugu-Ultra leads
SuperGPQASource 69.2%Not comparable
MMLU-ProSource 87.1%Not comparable
MMLU-Pro (Arcee)Source 87.1%Not comparable
HLESource 30.1%Not comparable
Artificial Analysis Intelligence IndexSource 35.4%Not comparable
AA-GPQA DiamondSource 87.9%Not comparable
AA-HLESource 29.4%Not comparable
AA-Omniscience IndexSource -8.1%Not comparable
AA-Omniscience AccuracySource 34.3%Not comparable
AA-Omniscience Hallucination RateSource 64.6%Not comparable
HLE w/o toolsSource 50%Not comparable
Math
BenchmarkKimi K2.5Sakana Fugu-UltraResult
AIME 2025Source 96.1%Not comparable
AIME26Source 95.8%Not comparable
AIME25 (Arcee)Source 96.3%Not comparable
HMMT Feb 2025Source 95.4%Not comparable
HMMT Nov 2025Source 91.1%Not comparable
HMMT Feb 2026Source 87.1%Not comparable
MMAnswerBenchSource 81.8%Not comparable
FrontierMath v2 (Tiers 1-3)Source 27.900%Not comparable
FrontierMath v2 (Tier 4)Source 4.200%Not comparable
Multilingual
BenchmarkKimi K2.5Sakana Fugu-UltraResult
MMLU-ProXSource 82.3%Not comparable
NOVA-63Source 56.0%Not comparable
MultimodalSakana Fugu-Ultra wins
BenchmarkKimi K2.5Sakana Fugu-UltraResult
MMMU-ProSource 78.5%Not comparable
Video-MMESource 87.4%Not comparable
MMVUSource 80.4%Not comparable
VideoMMMUSource 86.6%Not comparable
AA-MMMU-ProSource 75.4%Not comparable
Design Arena WebsiteSource 1279Not comparable
CharXivSource 86.6%Not comparable
Inst. Following
BenchmarkKimi K2.5Sakana Fugu-UltraResult
IFEvalSource 93.9%Not comparable
AA-IFBenchSource 70.2%Not comparable
Frequently Asked Questions (6)

Which is better, Kimi K2.5 or Sakana Fugu-Ultra?

Kimi K2.5 and Sakana Fugu-Ultra 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, Kimi K2.5 or Sakana Fugu-Ultra?

Sakana Fugu-Ultra has the edge for knowledge tasks in this comparison, averaging 95.5 versus 56.9. Inside this category, GPQA is the benchmark that creates the most daylight between them.

Which is better for coding, Kimi K2.5 or Sakana Fugu-Ultra?

Sakana Fugu-Ultra has the edge for coding in this comparison, averaging 64.5 versus 59.4. Inside this category, SWE-bench Pro is the benchmark that creates the most daylight between them.

Which is better for reasoning, Kimi K2.5 or Sakana Fugu-Ultra?

Sakana Fugu-Ultra has the edge for reasoning in this comparison, averaging 93.6 versus 61. Kimi K2.5 stays close enough that the answer can still flip depending on your workload.

Which is better for agentic tasks, Kimi K2.5 or Sakana Fugu-Ultra?

Sakana Fugu-Ultra has the edge for agentic tasks in this comparison, averaging 82.1 versus 55. Inside this category, Terminal-Bench 2.0 is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, Kimi K2.5 or Sakana Fugu-Ultra?

Sakana Fugu-Ultra has the edge for multimodal and grounded tasks in this comparison, averaging 86.6 versus 78.5. Kimi K2.5 stays close enough that the answer can still flip depending on your workload.

Self-host vs API cost

Estimates at 50,000 req/day · 1000 tokens/req average.

Kimi K2.5
API / mo$2,700
Self-host / mo$5,221
Break-even132M/day
Sakana Fugu-Ultra
API / mo$0
Self-host / moNot listed
Break-even
Proprietary model — self-hosting not applicable.
Model the full break-even

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.