Choose AionLabs: Aion 3.5 if…
- cost matters at scale — it is 67% cheaper on a typical 3:1 input/output mix ($345 vs $975 a month for 50,000 support chats)
- you want the newer model — released 2026-09-23, 3 months after the other
AionLabs: Aion 3.5 vs Sakana: Fugu Ultra
Neither wins outright: AionLabs: Aion 3.5 is the pick when cost matters at scale, Sakana: Fugu Ultra when you need to fit long inputs in one request.
Key numbers
| Spec | AionLabs: Aion 3.5 | Sakana: Fugu Ultra |
|---|---|---|
| Provider | Aion Labs | Sakana |
| Input price (per 1M tokens) | $3.00 | $5.00 |
| Output price (per 1M tokens) | $6.00 | $30.00 |
| Blended price (3:1 input/output) | $3.75 | $11.25 |
| Cheaper than … of all models we track | 17% | 5% |
| Context window | 262K tokens (~393 pages) | 1M tokens (~1,500 pages) |
| Larger context than … of all models | 47% | 69% |
| Max output per response | 33K tokens (~24,600 words) | 128K tokens (~96,000 words) |
| Input types | text | text, image |
| Output types | text | text |
| Free variant | No | No |
| Listed since | 2026-09-23 | 2026-06-24 |
Advertisement
What it costs for real workloads
Monthly cost at list price for five common workloads. Token counts per unit are stated so you can scale them to your own traffic. A workload that does not fit a model's context window is marked rather than priced.
| Workload (per month) | AionLabs: Aion 3.5 | Sakana: Fugu Ultra | Difference |
|---|---|---|---|
| Customer-support chatbot 50,000 conversations — each a 6-turn conversation (1,500 in / 400 out tokens) | $345 | $975 | AionLabs: Aion 3.5 saves $630 |
| RAG search over documents 20,000 questions — each a question answered from ~8 retrieved passages (6,000 in / 500 out tokens) | $420 | $900 | AionLabs: Aion 3.5 saves $480 |
| Coding / tool-using agent 2,000 tasks — each a multi-step task with tool calls and file context (60,000 in / 6,000 out tokens) | $432 | $960 | AionLabs: Aion 3.5 saves $528 |
| Batch summarisation 10,000 documents — each a 6,000-word report summarised to a page (8,000 in / 600 out tokens) | $276 | $580 | AionLabs: Aion 3.5 saves $304 |
| Long-document analysis 500 documents — each a 150-page contract or codebase read in one pass (100,000 in / 2,000 out tokens) | $156 | $280 | AionLabs: Aion 3.5 saves $124 |
Capabilities side by side
| Capability | AionLabs: Aion 3.5 | Sakana: Fugu Ultra | What it lets you do |
|---|---|---|---|
| Tool / function calling | ✓ Yes | ✓ Yes | call your APIs and run agent loops |
| Structured output (JSON schema) | ✓ Yes | ✓ Yes | return JSON that validates against your schema |
| Reasoning / thinking mode | ✓ Yes | ✓ Yes | spend extra tokens thinking before answering hard problems |
| Image input | — | ✓ Yes | read screenshots, charts and scanned pages |
| Audio input | — | — | take speech or audio directly |
| File / PDF input | — | — | accept documents without your own parsing step |
| Image output | — | — | generate images, not just text |
| Built-in web search | — | ✓ Yes | answer from live web results |
| Seed (reproducible sampling) | — | — | repeat a generation for tests and evals |
| Log-probabilities | — | — | read token confidence for classification and scoring |
From each model's published parameters and input/output types. “—” means not advertised, not proven absent.
Price history
- AionLabs: Aion 3.5: price unchanged since we started tracking it on 2026-09-24.
- Sakana: Fugu Ultra: price unchanged since we started tracking it on 2026-08-06.
Which should you choose?
Choose Sakana: Fugu Ultra if…
- you need to fit long inputs in one request — 1M tokens is roughly 1,500 pages, against 393
- you need image input — to read screenshots, charts and scanned pages; AionLabs: Aion 3.5 does not advertise it
- you need built-in web search — to answer from live web results; AionLabs: Aion 3.5 does not advertise it
- you generate long outputs in one go — up to 128K tokens (about 96,000 words) per response
Frequently asked questions
Is AionLabs: Aion 3.5 better than Sakana: Fugu Ultra?
Neither wins outright: AionLabs: Aion 3.5 is the pick when cost matters at scale, Sakana: Fugu Ultra when you need to fit long inputs in one request. In short: AionLabs: Aion 3.5 when cost matters at scale, you want the newer model; Sakana: Fugu Ultra when you need to fit long inputs in one request, you need image input, you need built-in web search, you generate long outputs in one go.
Which is cheaper, AionLabs: Aion 3.5 or Sakana: Fugu Ultra?
AionLabs: Aion 3.5 is cheaper: $3.75 vs $11.25 per 1M tokens at a typical 3:1 input/output mix, 67% less. For 50,000 support conversations a month that is $345 against $975.
How much do AionLabs: Aion 3.5 and Sakana: Fugu Ultra cost per million tokens?
AionLabs: Aion 3.5 costs $3.00 per 1M input tokens and $6.00 per 1M output tokens; Sakana: Fugu Ultra costs $5.00 input and $30.00 output. Output tokens are billed separately, so long answers shift the gap.
Which has the bigger context window, AionLabs: Aion 3.5 or Sakana: Fugu Ultra?
Sakana: Fugu Ultra: 1M tokens, about 1,500 pages, against 262K (about 393 pages) for AionLabs: Aion 3.5.
Do AionLabs: Aion 3.5 and Sakana: Fugu Ultra support function calling and JSON output?
Tool calling — AionLabs: Aion 3.5: yes, Sakana: Fugu Ultra: yes. JSON-schema structured output — AionLabs: Aion 3.5: yes, Sakana: Fugu Ultra: yes.
Can AionLabs: Aion 3.5 or Sakana: Fugu Ultra read images?
Only Sakana: Fugu Ultra accepts image input; AionLabs: Aion 3.5 is text-only for input.
Is AionLabs: Aion 3.5 or Sakana: Fugu Ultra free to use?
No free variant is listed for either; both are pay-per-token.
What is the maximum output length of AionLabs: Aion 3.5 and Sakana: Fugu Ultra?
AionLabs: Aion 3.5 can write up to 33K tokens (about 24,600 words) per response; Sakana: Fugu Ultra up to 128K (about 96,000 words).
When were AionLabs: Aion 3.5 and Sakana: Fugu Ultra released?
AionLabs: Aion 3.5 was listed on 2026-09-23; Sakana: Fugu Ultra on 2026-06-24.
About the two models
AionLabs: Aion 3.5
Aion 3.5 is a multi-model roleplaying and storytelling system from AionLabs, built on the GLM family of models. It uses a collaborative generation process in which multiple specialized models each...
Full AionLabs: Aion 3.5 specs →Sakana: Fugu Ultra
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...
Full Sakana: Fugu Ultra specs →