Fugu benchmarks, price, and specs | Tokens or Towers

AA Intelligence: —
Price: —
Output speed: —
Context window: —

Coding

SWE-bench Pro#15 of 31
59%vendormini-swe-agentSakana AI: Fugu v1 benchmark table ↗
Leader: Claude Opus 5.5 at 89.9%
Terminal-Bench 2.x#23 of 103
80.2%vendorTerminal-Bench 2.1Sakana AI: Fugu v1 benchmark table ↗
Leader: Claude Fable 5.1 at 91.4%
LiveCodeBench#1 of 56
92.9%vendorSakana AI: Fugu v1 benchmark table ↗
Leader: Fugu at 92.9%
SciCode#5 of 94
60.1%vendorSakana AI: Fugu v1 benchmark table ↗
Leader: Claude Opus 5.5 at 66.9%

Reasoning and knowledge

Humanity’s Last Exam#14 of 147
47.2%vendorSakana AI: Fugu v1 benchmark table ↗
Leader: Claude Opus 5.5 at 61.4%
GPQA Diamond#4 of 142
95.5%vendorSakana AI: Fugu v1 benchmark table ↗
Leader: GPT-6 Astra at 96.1%

Vision and long context

AA Long Context Reasoning#60 of 139
74.7%vendor"Long Context Reasoning"Sakana AI: Fugu v1 benchmark table ↗
Leader: Kimi K3 at 88.7%

Other published results

Result	Score	Reported by	Source
LiveCodeBench Pro	87.8%	vendor	Sakana AI: Fugu v1 benchmark table ↗
CharXiv Reasoning	85.1%	vendor	Sakana AI: Fugu v1 benchmark table ↗
tau3-Bench Banking	21.7%	vendor	Sakana AI: Fugu v1 benchmark table ↗
MRCRv2	86.6%	vendor	Sakana AI: Fugu v1 benchmark table ↗