Sakana

Fugu Ultra

Generally availableProprietary1M contextReasoning

Summary

Fugu Ultra is a proprietary language model from Sakana, released on 15 Jun 2026. It accepts text and images and generates text, with a 1M-token context window and up to 128K output tokens per response. As of 4 Oct 2026, the lowest listed API price is $5 per million input tokens and $30 per million output tokens (OpenRouter), across 2 providers we track.

Quality-first multi-agent model for hard research, analysis, and competitions

Last verified 4 Oct 2026. Every figure on this page links to its source.

Specifications

DeveloperSakana
Release date15 Jun 2026 [models.dev]
StatusGenerally available [models.dev]
WeightsProprietary (API only) [models.dev]
Context window1M (1,000,000 tokens) [models.dev]
Max output128K tokens [models.dev]
InputText and images [models.dev]
OutputText [models.dev]
Reasoning modeYes [models.dev]
Tool callingYes [models.dev]
Structured outputYes [models.dev]

Fugu Ultra API pricing by provider

USD per million tokens. Sorted by input price.

ProviderInputOutputCached inputContextSource
OpenRouter
sakana/fugu-ultra
$5$30$0.501MOpenRouter, 4 Oct 2026
Vercel AI Gateway
sakana/fugu-ultra
$5$30$0.501Mmodels.dev, 4 Oct 2026

Frequently asked questions

What is the context window of Fugu Ultra?

Fugu Ultra has a context window of 1,000,000 tokens (1M) and can generate up to 128,000 tokens in a single response, according to models.dev.

How much does the Fugu Ultra API cost?

Fugu Ultra is listed from $5 per million input tokens and $30 per million output tokens (OpenRouter). Prices last verified 4 Oct 2026.

Is Fugu Ultra open source?

No. Fugu Ultra is a proprietary model; Sakana has not released its weights. It is available through APIs including OpenRouter, Vercel AI Gateway.

When was Fugu Ultra released?

Sakana released Fugu Ultra on 15 Jun 2026, according to models.dev.

Which providers offer Fugu Ultra?

We track Fugu Ultra on 2 providers: OpenRouter, Vercel AI Gateway.

Change history

No changes detected since we started tracking this model. We re-check every source daily.

Sources