Simple pricing.
Real AI.

An iToken (Inference Token) is the unit of compute on premium model runs and high-volume jobs. The base membership covers everyday usage; iTokens let you tap heavier models or burst capacity for specific tasks. Pay-as-you-go at $1/token, or buy a 1,000-pack for $900 (10% off).

Your membership comes with one Personal AI Agent — a general-purpose assistant. The Agent Trainer is a paid service ($50–$500) where our team customises an agent for a specific job: training data curation, prompt-engineering, evaluation suite, optionally a fine-tuned LoRA adapter for the deepest tier. Turnaround: 24 hours to 2 weeks depending on depth.

Yes. Monthly memberships cancel at the end of the current period. Annual memberships are paid upfront for 12 months and cancellation takes effect at renewal. No questions, no retention calls.

All conversations and knowledge-base content are scoped to your member ID and are never shared across members. Storage is in the GME-controlled environment under our Singapore / Malaysia / Hong Kong entities. We do not train any base model on your private content. Your feedback signals (👍 / 👎) are anonymised and used only to improve prompts and routing — never the data itself.

The platform is fully multilingual across English, 中文, 日本語, 한국어, ไทย, Bahasa Melayu, Bahasa Indonesia, Español, Français, and العربية — both for sign-in and for conversation. Models are routed by language at inference time.

Yes — for enterprise tiers ($5k+/month). We deploy the entire stack (vector DB, MCP, skills, vLLM with Llama 3.3 / Qwen 2.5) inside your environment. Air-gapped variant available. Contact us for a sandbox proposal.