Launch Into The AI Universe

You get 1,000,000 platform tokens and up to 100 requests/day. You can upgrade anytime or top up with token packages.

Current catalog includes 209 active models from 12 providers.

Create a Bearer token in API Tokens. Free-clone web models are blocked in API; use paid-available models with platform tokens or your own provider key.

Yes. Each API token can have monthly request and monthly platform-token limits. Exceeding either limit returns HTTP 429.

Remote Control lets you connect SSH servers, test connection, and run controlled automation scenarios through remote agents.

Yes. You can schedule messages, manage reminder flows, share chats by public link, and use live streaming with tool events.

AI Studio supports image, video, and audio generation with model-specific controls, board workflow, and export actions.

Use Team Workspaces with shared chats, files, notes, todos, reminders, and team messages in one project context.

AIRocket is a unified AI platform: chat, agents, catalog, workspaces and API. Start with Chat, select a model, send one clear task, then review usage in Usage.

Use lightweight models for speed/cost, reasoning models for deep logic, code-oriented models for programming, and vision models for screenshots/images.

THK means thinking/reasoning support, TLS means tools/functions support, RT means realtime mode. Choose model tags based on required features.

Provider is the API service (OpenAI, Anthropic, etc.). Model is a specific engine inside that provider. One provider can host many models.

Add your key if you need provider-side billing, your own limits, or finer access control. Keys are encrypted but must never be shared in chats/tickets.

Common reasons: model is not API-allowed for your limits/plan, or provider key is inactive/invalid. Check API token, model id, provider key and access rules.

A single agent executes tasks directly. An orchestrator (multi-agent) splits complex work into steps and delegates to sub-agents, which is usually better for multi-step tasks.

Set guardrails: Max Steps, Max Sub-agent Calls, Max Run Seconds, and Budget USD. Keep sub-agents narrowly specialized (Planner/Engineer/QA).

Templates are reusable user prompts, Skills are reusable capability blocks, and System Prompts are role-level rules. Save successful patterns for team reuse.

Use one workspace per project/client. Keep decisions in notes, tasks in todos/reminders, communication in team chat, and upload only relevant files.

Tokens are text units used for limits and billing. Long prompts and long chat history increase token usage and cost.

Use: context → goal → constraints → output format → example. If quality is weak, narrow scope, add acceptance criteria, and request step-by-step output.

This may be hallucination. For critical tasks, require verification, cross-check with multiple models, and add explicit validation steps.

Open Support: review guide and FAQ first, then create a ticket with reproduction steps, expected behavior, and actual error details.