AMIYO

think beyond · build 37no provider yet

0 providers 0 agents

THINK BEYOND.

Connect your AI provider

AMIYO needs an AI provider to start. Add one below — OpenRouter, OpenAI, Ollama, a free provider that needs no key, or anything OpenAI/Anthropic/Gemini-compatible.

📋 Paste a curl command, .env line, or CLI config instead

Provider Name

Provider API

Model Name

✓ connected

Model:

AMIYO is ready.

My AI Providers

Connect, test and route between your AI providers.

My Agents

One model, or a team across providers. Turn on 🤖 Agentic and an agent plans and works step by step.

◆ Providers & routing

Translate

Uses the providers you've connected, with your own API key. Nothing else is involved.

From

Translate with

Read aloud with

Voice

Speed

Read the translation aloud as soon as it's ready Also show it in English letters (how to pronounce it)

Export config

Your providers, agents and routing choices as a JSON file. API keys are secrets — choose the version without them if the file will be shared, emailed or uploaded anywhere.

Testing providers

Each provider is tried one at a time.

Chats

Every conversation is saved on this device. Tap one to reopen it.

Saved

Messages you've starred with “save”, kept independent of the chat they came from.

Media Library

Every image, video and sound AMIYO has generated on this device, in one place.

Browse models

Tap ★ to set a model as default, or + to add it without switching.

Free only

Modality, origin, size and price are guessed from each model's name — not always exact.

Connect AI Provider

Minimum required: name, API key, model. Everything else is optional.

📋 Paste a curl command, .env line, or CLI config to autofill

Provider Name

Provider API

Model Name

Fill in name + API key (and endpoint, if this provider isn't a known preset), then tap this to pull its live model list.

Free only

Advanced Settings

API Endpoint

Authentication

Request Format

Header Name

Request Body Template (JSON)

Use {{message}} and {{model}} as placeholders.

Response Path

Dot path to the reply text in the provider's JSON response.

Image generation (optional)

Image models

Comma-separated. Leave blank to use AMIYO's defaults for OpenAI, OpenRouter, Gemini and Higgsfield.

Image format

Image endpoint

Video generation (optional)

Video models

Comma-separated. Leave blank to use AMIYO's defaults for OpenAI (Sora), Gemini (Veo), OpenRouter and Higgsfield.

Video format

Video endpoint

Sound generation (optional)

Sound models

Comma-separated. Covers speech, sound effects and song/music models. On ElevenLabs, use the names sound-effects and music for those two. Leave blank for AMIYO's defaults on OpenAI and ElevenLabs.

Sound format

Sound endpoint

Default voice / voice ID

Used when a message doesn't pick a named voice — mainly for ElevenLabs, which addresses voices by ID.

Custom Headers (JSON)

Timeout (seconds)

Retry attempts

Enable streaming

Build Agent

An agent is a saved assistant: one model, or a team of models from different providers that coordinate on every message.

Agent Name

Icon

Description (optional)

System Prompt (shared by every model in the agent)

How the models work together

Review rounds

🤖 Agentic mode — plans, uses tools and works step by step until the goal is done

Tools this agent may use

Max steps per task

Every step is one request to the agent's model, so long tasks use more of your provider credits. For a team agent, the lead model does the thinking and the other models take over if it fails.

Advanced Settings

Temperature (0–2, optional)

Max tokens (optional)

Left blank, requests use the provider's own defaults. Not all formats honor these.