Quality-first multi-agent model for hard research, analysis, and competitions
Fugu Ultra by sakana is offered by 12 providers on this page. Its official list price is $5.00 per 1M input tokens and $30.00 per 1M output tokens.
The context window is 1,000,000 tokens at the reference host, but hosts report different limits, from 1,000,000 to 1,048,576, so the usable window depends on the provider you pick. It supports reasoning, tool use, and structured output. Accepted input modalities are Text and Image.
| Provider | Tier | Input | Output | Cache read | Cache write | Context | Output limit | Status |
|---|---|---|---|---|---|---|---|---|
| NanoGPT | Gateway | $5.00 | $30.00 | $0.50 | — | 1,000,000 | 16,384 | |
| Kilo Gateway | Gateway | $5.00 | $30.00 | $0.50 | — | 1,000,000 | 128,000 | |
| OpenRouter | Gateway | $5.00 tiered >272K: $10.00 | $30.00 | $0.50 | — | 1,000,000 | 128,000 | |
| LLMTR | Gateway | $5.00 | $30.00 | — | — | 1,000,000 | 1,000,000 | |
| Vercel AI Gateway | Cloud | $5.00 tiered >272K: $10.00 | $30.00 | $0.50 | — | 1,000,000 | 1,000,000 | |
| Sakana AIOfficial | First-party | $5.00 tiered >272K: $10.00 | $30.00 | $0.50 | — | 1,000,000 | 1,000,000 | |
| Sakana AIOfficial fugu-ultra-20260615 | First-party | $5.00 tiered >272K: $10.00 | $30.00 | $0.50 | — | 1,000,000 | 1,000,000 | |
| Merge Gateway | Gateway | $5.00 | $30.00 | $0.50 | — | 1,000,000 | 128,000 | |
| Pioneer | Gateway | $5.00 | $30.00 | $0.50 | $5.00 | 1,000,000 | 131,072 | |
| Requesty | Gateway | $5.00 tiered >272K: $10.00 | $30.00 | $0.50 | — | 1,048,576 ⚠ | 131,072 | |
| DevPass (LLM Gateway) | Gateway | $5.00 | $30.00 | $0.50 | — | 1,000,000 | 1,000,000 | |
| LLM Gateway | Gateway | $5.00 | $30.00 | $0.50 | — | 1,000,000 | 1,000,000 |
Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.
Interleaved thinking (reasoning between tool calls) is declared by 1 of 12 providers.
| Name | Conditions | Score | Metric | Source |
|---|---|---|---|---|
| SWE Bench Pro | — | 73.7 | — | Source ↗ |
| Terminal Bench 2.1 | — | 82.1 | — | Source ↗ |
| LiveCodeBench | — | 93.2 | — | Source ↗ |
| LiveCodeBench Pro | — | 90.8 | — | Source ↗ |
| Humanity’s Last Exam | — | 50 | — | Source ↗ |
| CharXiv Reasoning | — | 86.6 | — | Source ↗ |
| GPQA Diamond | — | 95.5 | — | Source ↗ |
| SciCode | — | 58.7 | — | Source ↗ |
| τ3 Banking | — | 20.6 | — | Source ↗ |
| Long Context Reasoning | — | 73.3 | — | Source ↗ |
| MRCRv2 | — | 93.6 | — | Source ↗ |
| CTI-REALM | — | 69.4 | — | Source ↗ |
The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.