Fugu vs Fugu Ultra
Fugu (2026) and Fugu Ultra (2026) are frontier-tier reasoning models from Sakana AI. Fugu ships a 1m-token context window, while Fugu Ultra ships a 1m-token context window. On SWE-bench Pro, Fugu Ultra leads by 14.7 pts. This comparison covers specs, pricing, API access, capabilities, benchmarks, input and output token costs, and production fit for coding and agent workloads. It focuses on practical selection signals rather than broad model-family marketing.
Fugu is safer overall; choose Fugu Ultra when vision-heavy evaluation matters.
Decision scorecard
Local evidence first| Signal | Fugu | Fugu Ultra |
|---|---|---|
| Best for | reasoning-heavy apps, multimodal apps, and tool-calling agents | reasoning-heavy apps, multimodal apps, and tool-calling agents |
| Decision fit | Coding, RAG, and Agents | Coding, RAG, and Agents |
| Context window | 1m | 1m |
| Cheapest output | - | $30/1M tokens |
| Provider routes | 1 tracked | 2 tracked |
| Shared benchmarks | 6 shared | SWE-bench Pro leader |
Decision tradeoffs
- Local decision data tags Fugu for Coding, RAG, and Agents.
- Fugu Ultra holds a shared-benchmark lead on SWE-bench Pro, ahead by 14.7 points.
- Fugu Ultra has broader tracked provider coverage for fallback and procurement flexibility.
- Local decision data tags Fugu Ultra for Coding, RAG, and Agents.
Monthly cost at traffic
Estimate token spend from the cheapest tracked input and output route or tier on this page.
Fugu
Unavailable
No complete token price in local provider data
Fugu Ultra
$11,500
Cheapest tracked route/tier: Sakana AI <=272K tokens
Cost delta unavailable until both models have sourced input and output token prices.
Switch friction
- Provider overlap exists on Sakana AI; start route-level A/B tests there.
- Provider overlap exists on Sakana AI; start route-level A/B tests there.
Specs
| Specification | ||
|---|---|---|
| Released | 2026-06-22 | 2026-06-22 |
| Context window | 1m | 1m |
| Parameters | — | — |
| Architecture | Composition of Experts | Composition of Experts |
| License | Proprietary | Proprietary |
| Openness | Proprietary | Proprietary |
| Commercial use | Commercial use: conditional | Commercial use: conditional |
| Knowledge cutoff | - | - |
Pricing and availability
| Pricing attribute | Fugu | Fugu Ultra |
|---|---|---|
| Input price | - |
|
| Output price | - |
|
| Providers |
Capabilities
| Capability | Fugu | Fugu Ultra |
|---|---|---|
| Vision | Yes | Yes |
| Multimodal | Yes | Yes |
| Reasoning | Yes | Yes |
| Function calling | Yes | Yes |
| Tool use | Yes | Yes |
| Structured outputs | Yes | Yes |
| Code execution | No | No |
| IDE integration | No | No |
| Computer use | No | No |
| Parallel agents | No | No |
Benchmarks
| Benchmark | Fugu | Fugu Ultra |
|---|---|---|
| SWE-bench Pro | 59.0 | 73.7 |
| Google-Proof Q&A | 95.5 | 95.5 |
| LiveCodeBench | 92.9 | 93.2 |
| Humanity's Last Exam | 47.2 | 50.0 |
| Terminal-Bench 2.1 | 80.2 | 82.1 |
| CharXiv | 85.1 | 86.6 |
Deep dive
On shared benchmark coverage, SWE-bench Pro has Fugu at 59 and Fugu Ultra at 73.7, with Fugu Ultra ahead by 14.7 points; Google-Proof Q&A has Fugu and Fugu Ultra tied at 95.5; LiveCodeBench has Fugu at 92.9 and Fugu Ultra at 93.2, with Fugu Ultra ahead by 0.3 points. The largest visible gap is 14.7 points on SWE-bench Pro, which matters most when that benchmark mirrors your workload. Treat isolated benchmark wins as directional, because provider routing, prompt style, and tool access can move real application results.
The capability footprint is close: both models cover vision, multimodal input, reasoning mode, function calling, and tool use. That makes context budget, benchmark fit, and provider maturity more important than a simple checklist. If your application depends on one integration detail, verify it against the provider route you plan to use, not just the base model listing.
Pricing coverage is uneven: Fugu has no token price sourced yet and Fugu Ultra has $5/1M input tokens. Provider availability is 1 tracked routes versus 2. Treat unknown pricing as an integration gap, then verify the route you will actually call before estimating production spend.
Choose Fugu when vision-heavy evaluation are central to the workload. Choose Fugu Ultra when vision-heavy evaluation and broader provider choice are more important. For production, rerun your own prompts through the exact provider, region, and tool stack you plan to ship.
FAQ
Which has a larger context window, Fugu or Fugu Ultra?
Fugu supports 1m tokens, while Fugu Ultra supports 1m tokens. That gap matters most for long documents, large codebases, retrieval-heavy agents, and conversations where earlier context must remain visible. Use this as a quick comparison signal, then confirm the provider-specific limits before committing to production.
Is Fugu or Fugu Ultra open source?
Fugu is listed under Proprietary. Fugu Ultra is listed under Proprietary. License labels affect whether you can self-host, redistribute weights, or rely only on hosted APIs, so confirm the upstream license before deployment.
Which is better for vision, Fugu or Fugu Ultra?
Both Fugu and Fugu Ultra expose vision. The better choice depends on benchmark fit, context budget, pricing, and whether your provider route exposes the same capability surface. Use this as a quick comparison signal, then confirm the provider-specific limits before committing to production.
Which is better for multimodal input, Fugu or Fugu Ultra?
Both Fugu and Fugu Ultra expose multimodal input. The better choice depends on benchmark fit, context budget, pricing, and whether your provider route exposes the same capability surface. Use this as a quick comparison signal, then confirm the provider-specific limits before committing to production.
Which is better for reasoning mode, Fugu or Fugu Ultra?
Both Fugu and Fugu Ultra expose reasoning mode. The better choice depends on benchmark fit, context budget, pricing, and whether your provider route exposes the same capability surface. Use this as a quick comparison signal, then confirm the provider-specific limits before committing to production.
Where can I run Fugu and Fugu Ultra?
Fugu is available on Sakana AI. Fugu Ultra is available on Sakana AI and Vercel AI Gateway. Provider coverage can affect latency, region availability, compliance posture, and fallback options. Use this as a quick comparison signal, then confirm the provider-specific limits before committing to production.
Continue comparing
Last reviewed: 2026-06-22. Data sourced from public model cards and provider documentation.