Every price,
in one place.
Every metered price on the platform. One credit is one penny — so a rate here is a rate in cents, and ten thousand credits is a hundred dollars of work. Nothing is bundled and nothing is a minimum: you pay for what you actually run, and for nothing you don’t. Back to pricing
AI models
Dollars per 1M tokens. Every model the platform can run, priced the same way — so you can pick the one that fits the job, and move the moment it stops fitting.
| Model | Input | Output |
|---|---|---|
| anthropic | ||
| Fable 5 | $12 | $60 |
| Haiku 4.5 | $1.2 | $6 |
| Haiku 5.5 | $0.12 | $0.6 |
| Opus 4.6 | $6 | $30 |
| Opus 4.7 | $6 | $30 |
| Opus 4.8 | $6 | $30 |
| Opus 5 | $6 | $30 |
| Opus 5.5 | $4.8 | $24 |
| Sonnet 4.6 | $3.6 | $18 |
| Sonnet 5 | $2.4 | $12 |
| Sonnet 5.5 | $2.4 | $12 |
| Gemini 3.1 Pro | $2.4 | $14.4 |
| Gemini 3.6 Flash | $1.8 | $9 |
| meta-selfhost | ||
| Llama 3.1 70B (dedicated) | $10 | $10 |
| moonshot | ||
| Kimi K2.7 Code | $1.14 | $4.8 |
| Kimi K3 | $3.6 | $18 |
| openai | ||
| Embeddings (3-large) | $0.156 | $0 |
| Embeddings (3-small) | $0.024 | $0 |
| GPT-5.6 Sol | $6 | $36 |
| zai | ||
| GLM-5.2 | $1.68 | $5.28 |
Everything else
Metered per unit of work. You are charged when something runs, and never otherwise.
Build & ship
Documents & media
Voice & avatar
Research & data
Secure & comply
Store & serve
Voice & video calls themselves aren’t metered — you only spend credits on transcription, recording, and AI on top. Representative rates; changes announced 30 days in advance.
Run out? Work pauses, you top up in one click — no overage invoice, no surprise bill, ever.