Models
Your keys, any model.
Ideyo calls the providers’ APIs directly. You attach your own keys, as many per provider as you need, choose a model for each stage, from the catalog below or from OpenRouter’s live list, and see the forecast before you run. Without a key, the demo model walks the whole pipeline for free, and managed purchases run on Ideyo’s keys.
The curated catalog. List prices in US dollars per million tokens, as published by each provider and checked on 6 October 2026. With your own key you pay these prices to the provider; Ideyo adds nothing.
| Model | Provider | Tier | Input | Output | Context | Web search |
|---|---|---|---|---|---|---|
| Claude Opus 5.5 | Anthropic | Frontier | $4.00 | $20.00 | 1M tokens | |
| Claude Sonnet 5.5 | Anthropic | Balanced | $2.00 | $10.00 | 1M tokens | |
| Claude Haiku 4.5 | Anthropic | Fast | $1.00 | $5.00 | 200k tokens | |
| Claude Fable 5.1 | Anthropic | Frontier | $10.00 | $50.00 | 1M tokens | |
| GPT-6 Astra | OpenAI | Frontier | $10.00 | $50.00 | 1.1M tokens | |
| GPT-6.1 Sol | OpenAI | Balanced | $2.00 | $10.00 | 1.1M tokens | |
| GPT-6 Luna | OpenAI | Fast | $0.10 | $0.50 | 1.1M tokens | |
| Gemini 3.1 Pro (preview) | Google AI Studio | Frontier | $2.00 | $12.00 | 1M tokens | |
| Gemini 3.8 Flash | Google AI Studio | Balanced | $0.75 | $3.75 | 1M tokens | |
| Gemini 3.5 Flash-Lite | Google AI Studio | Fast | $0.30 | $2.50 | 1M tokens | |
| Mistral Medium 3.5 | Mistral AI | Frontier | $1.50 | $7.50 | 256k tokens | |
| Mistral Small 4 | Mistral AI | Fast | $0.15 | $0.60 | 256k tokens | |
| DeepSeek V4 Pro | DeepSeek | Balanced | $1.32 | $3.96 | 1M tokens | |
| DeepSeek V4.1 Flash | DeepSeek | Fast | $0.30 | $1.20 | 1M tokens | |
| Kimi K3 (OpenRouter) | OpenRouter | Balanced | $0.86 | $14.00 | 1M tokens |
And about 300 more through OpenRouter. With an OpenRouter key, the model picker also lists OpenRouter’s live catalog: every model that can call tools, reads and writes text and has a context window large enough for a stage, with its current price. The list is refreshed every six hours and searchable by name or vendor, and the forecast uses the same prices.
Routing presets. Start from a preset, then change any line. Ideyo only offers the models you hold a valid key for, and the forecast follows every change: Fit my budget moves the stages that cost most to cheaper models until the expected total fits your cap.
Balanced
The strongest model where judgement matters (Discover, Design, Quality), a mid-range one for Specify and Build, a fast one to ship.
Frontier
The strongest model you have a key for, at every stage. The best results, at the highest cost.
Economy
The fastest, cheapest model you have a key for, at every stage. Good for a first look at many ideas.
Demo
Ideyo’s scripted demo model. No key and no cost: it walks a realistic project through every stage so you can see how Ideyo works.
Your keys, and how they are kept.
- Add several keys per provider and name them (personal, a client, a test budget). One is the default; any stage of a project can be pinned to another.
- Keys are encrypted at rest with AES-256-GCM under a master key that never leaves the server, and decrypted only for the call that needs them.
- When you add a key, Ideyo validates it with one cheap call. After that, only its last four characters are ever displayed, with when it was last used.
- Your keys are used for your projects only. They are never pooled, shared or used for anyone else.
- Delete a key at any time. Deleting your account deletes all of them.
No key? Buy the deliverable. A Study ($49) or a Managed project ($390) runs on Ideyo’s own API keys, under the providers’ commercial terms. You buy a deliverable, not tokens: the inference Ideyo pays for is capped ($120 for a managed project), and going beyond it needs your agreement.
Where to get a key. Each provider issues keys from its own console. Create one, paste it into Ideyo’s settings, and it is checked straight away.
- Anthropicplatform.claude.com
- OpenAIplatform.openai.com
- Google AI Studioaistudio.google.com
- Mistral AIconsole.mistral.ai
- DeepSeekplatform.deepseek.com
- OpenRouteropenrouter.ai