Sakana Fugu
Multi-agent model in AI tools
Recommended for multi-model routing that aims for frontier-level results at lower cost. Main caveat: it feels slow and uses up usage limits fast.
Recommended for multi-model routing that aims for frontier-level results at lower cost. Main caveat: it feels slow and uses up usage limits fast.
33 people gave an opinion
- recommend it
- 12
- criticize it
- 21
- approval share
- 36%
Recommended in 3 of 13 months
- November 2025: not recommended
- December 2025: not recommended
- January 2026: not recommended
- February 2026: not recommended
- March 2026: not recommended
- April 2026: not recommended
- May 2026: not recommended
- June 2026: recommended
- July 2026: not recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: not recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 12 people recommend itNeeds 10 or more, met.
- 0.5 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 3 of 13 monthsNeeds 4 or more, not met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Multi-model routing yields better results
“Clever routing systems also seem to indicate frontier or frontier-beating capabilities at reduced cost, such as Fugu.”
Comment from 6 Aug 2026 in “Beating GPT-5.6 Sol on retrieval with 100x cheaper open models” “being able to pick from a panoply of models will actually yield much better results in the long run”
Comment from 22 Jun 2026 in “Sakana Fugu” Strong benchmarks and reviews
“it's a genuine accomplishment to hit those benchmarks and offer a reasonable plan”
Comment from 22 Jun 2026 in “Sakana Fugu” “Beats GPT-5.5 xhigh on complex tasks.”
Comment from 23 Jun 2026 in “Sakana Fugu” “the reviews were quite strong”
Comment from 22 Jun 2026 in “Sakana Fugu”
What people warn about
Each mark is one comment
Priced like frontier models
“But it's priced the same as frontier models. Why do I not directly pay for frontier models?”
Comment from 22 Jun 2026 in “Sakana Fugu” “seems super slow and expensive, and everyone i've talked to who tried it gave up”
Comment from 1 Jul 2026 on Hacker News “you pay $200/month to Sakana to coordinate it all, because why not.”
Comment from 22 Jun 2026 in “Sakana Fugu” - Comment from 22 Jun 2026 in “Sakana Fugu”
Underwhelming results for many users
“Not that impressed with the results I got from it however if I'm being honest.”
Comment from 27 Jun 2026 in “Show HN: Smart model routing directly in Claude, Codex and Cursor” “Missed half the stuff the other half was outdated/ didn't verify.”
Comment from 27 Jun 2026 in “Asian AI startups launch Mythos-like models” “Seems kinda underwhelming considering they raised like $400M.”
Comment from 22 Jun 2026 in “Sakana Fugu” - Comment from 23 Jun 2026 in “A New Competitor for Fable 5 and Mythos Preview: Sakana's Fugu Ultra Model”
Feels slow in practice
“the result was worse than Opus, incredibly slow, and I ended up exhausting the new 5 hour window”
Comment from 27 Jun 2026 in “Asian AI startups launch Mythos-like models” “seems super slow and expensive, and everyone i've talked to who tried it gave up”
Comment from 1 Jul 2026 on Hacker News “The 5-hour limit runs out surprisingly fast. Quality is okay but it feels slow”
Comment from 22 Jun 2026 in “Sakana Fugu”
What people said
35 of the 38 comments that mention it, newest first. Each links to the original comment.
“it seems to produce results of extremely varying quality”
“Clever routing systems also seem to indicate frontier or frontier-beating capabilities at reduced cost, such as Fugu.”
“Fugu is kind of a black box in terms of what model is used for what”
“seems super slow and expensive, and everyone i've talked to who tried it gave up”
“I canceled my test subscription in two days.”
“They just use multiple existing SaaS models from OpenAI, Anthropic etc in background”
“you're paying api markup twice”
“20$ is not enough to complete a full workflow”
“I experienced the same exact thing”
“Missed half the stuff the other half was outdated/ didn't verify.”
“the result was worse than Opus, incredibly slow, and I ended up exhausting the new 5 hour window”
“Not that impressed with the results I got from it however if I'm being honest.”
“It gives me a small army that can break down and slice, dice the problem in different ways”
“Beats GPT-5.5 xhigh on complex tasks.”
“it didn't impress me enough to put more into it”
“He makes some very good arguments for them.”
“a questionably better black box for slower service and more money”
“it had the same sycophantic tendencies as most LLMs these days”
“The 5-hour limit runs out surprisingly fast. Quality is okay but it feels slow”
“so we put agents in yo agents so you can burn tokens while you burn tokens.”
“it's a genuine accomplishment to hit those benchmarks and offer a reasonable plan”
“being able to pick from a panoply of models will actually yield much better results in the long run”
“perhaps this particular product is not well thought out”
“the reviews were quite strong”
“seems like an easy thing for a frontier lab to make obsolete”
“I wouldn't personally pay for this service”
“But it's priced the same as frontier models. Why do I not directly pay for frontier models?”
“as they are involved in military contracts”
“rather than commercial models that are only availble via an API”
“usage felt more efficient token wise. Switching models every time is annoying.”
“They'll be incentivized for your success, not token-maximizing for their investors.”
“you pay $200/month to Sakana to coordinate it all, because why not.”
“I thought sakana.ai was doing cooler stuff than this”
“Their research around building a domain specific model is pretty cool”
“Seems kinda underwhelming considering they raised like $400M.”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.