uttapen

Fugu Ultra API: toman pricing and code

sakana/fugu-ultra

visiontoolsreasoningjson
Input · per 1M tokens
1,450,625 toman
Output · per 1M tokens
8,703,750 toman
One 1,000-word request ≈
13,201 toman

You are billed for the usage the request actually reported. Prices follow the market. Cached input: 145,063 toman / 1M. Web search: 2,901.25 toman / request.Pricing and top-ups

Fugu Ultra example: reading an image into JSON

The example is picked from this model's own capabilities. Drop in your key and it runs as is.

# tip: a data URI works too — base64 the file and prefix it with data:image/jpeg;base64,
curl https://api.uttapen.ir/v1/chat/completions \
  -H "Authorization: Bearer sk-up-…" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "sakana/fugu-ultra",
    "messages": [{
      "role": "user",
      "content": [
        {"type": "text", "text": "Return only the invoice number and the total, as JSON."},
        {"type": "image_url", "image_url": {"url": "https://example.com/factor.jpg"}}
      ]
    }]
  }'

What is Fugu Ultra good for?

Sakana publishes this model; we expose it under the id "sakana/fugu-ultra". Its context window is 1,000,000 tokens, roughly 750k English words in one request. A single response can run to 128,000 tokens. On price it sits in the "expensive" band — cheaper than 381 and dearer than 25 of the other paid models in the catalogue.

What it can do beyond plain text: it reads images directly, which makes it a real option for invoices, forms and screenshots; it supports tool calling, so it can invoke your own functions with valid arguments; it returns schema-valid JSON through response_format, ready to hand to your code; it has a reasoning mode that pays off on multi-step problems, maths and debugging. All of it works through the standard parameters of the official OpenAI SDK — no custom client, no wrapper. Keep in mind that reasoning tokens are output tokens and do appear on the bill.

Pricing is 1,450,625 toman per 1M input tokens and 8,703,750 per 1M output tokens. A 1,000-word round trip on Fugu Ultra lands near 13,201 toman. It supports cached input: repeated context is billed at 145,063 toman per 1M, which matters a lot if your system prompt is long. You are always charged for the usage the request actually reported, never for the estimate, and a failed request costs nothing.

The closest alternative with the same capabilities from a different provider is GPT-4 Turbo: Fugu Ultra works out roughly 1.1× cheaper, and its context window is larger. Both run on the same key and the same code, so trying the other one is a single string change.

To make the figure concrete: 100,000 toman of credit buys roughly 8 thousand-word requests on Fugu Ultra, and every 1,000 toman is about 76 words of round trip. A job with one million input tokens and one million output tokens comes to 10,154,375 toman in total. Filling this model's 1,000,000-token window costs 1,450,625 toman on the input side alone, which is the real reason to keep conversation history short.

Its nearest relative in the catalogue is "sakana/sakana-namazu", and choosing between those two is where most people hesitate. On a thousand-word request this variant works out 7.1× dearer (13,201 against 1,867 toman). The context windows differ too: 1,000,000 against 262,144 tokens. Maximum answer length differs as well: 128,000 against 65,536 tokens.

Sakana has 2 models in our catalogue; the cheapest is Sakana Namazu at 1,867 toman per thousand words and the dearest Fugu Ultra at 13,201. Among the less common parameters it accepts reasoning_effort, web_search_options — all through the standard request body. It does not support max_tokens, response_format, temperature, top_p, which most models here do, so test before switching if your code relies on them. By context size the nearest option from another provider is Nova 2 Lite at 1,000,000 tokens.

Good fits: work where the quality of the answer matters more than its cost, logic puzzles and code review, pulling text and fields out of images, agents that reach out to APIs and databases, extracting data against a fixed schema, analysing a long document or codebase in one request.

Provider's own description

Fugu Ultra is the higher-performance model in Sakana AI's Fugu family. Rather than a single monolithic model, Fugu is a learned multi-agent orchestration system: a language model trained to route...

Three real jobs, priced on this model

Each figure is derived from the prices above and moves when they do.

JobTokensCost
One chat turn with a medium history1,500 in + 400 out5,657 toman
Summarising a ten-page document4,000 in + 600 out11,025 toman
Classifying a thousand short rows120,000 in + 20,000 out348,150 toman

Frequently asked

How do I call Fugu Ultra from Iran?
Sign up with your mobile number, top the wallet up in toman, create an API key, then in the official OpenAI SDK point base_url at https://api.uttapen.ir/v1 and set model to "sakana/fugu-ultra". Nothing else in your code changes.
What does Fugu Ultra cost in toman?
1,450,625 toman per 1M input tokens and 8,703,750 toman per 1M output tokens; a 1,000-word request is around 13,201 toman. You pay for the usage that request actually reported, and a failed request costs nothing.
How much input does Fugu Ultra take?
Up to 1,000,000 tokens per request, roughly 750k words. A single answer can reach 128,000 tokens.
Does Fugu Ultra support streaming and tool calling?
Streaming (stream=true) works on every model here. This one supports tool calling in the standard OpenAI shape. Image input is accepted through image_url, as a data URI or a public URL. Structured output through response_format works too.