Pricing
Last Updated: April 30, 2026
Hosted pricing comes down to the product you run and whether you want standard or priority service. The agent pays via x402 on Solana in USDC or USDT at dollar-equivalent rates. Hosted Memory uses 7-day retention by default, and 30-day retention adds $0.50 per request.
How Pricing Works
Pick the hosted product first. Then decide whether to stay on standard service or pay the priority rate when faster admission matters.
Product
Hosted Chat
Stateless traffic and one-off requests.
- Input
- $5 / 1M tokens
- Output
- $30 / 1M tokens
- Retention
- None
Product
Hosted Memory
Persistent agents and workflows with 7-day work retention.
- Input
- $10 / 1M tokens
- Output
- $60 / 1M tokens
- Retention
- 7 days included
Product
Hosted Memory
Persistent agents and workflows with 30-day project retention.
- Input
- $10 / 1M tokens
- Output
- $60 / 1M tokens
- Retention
- +$0.50 / request for 30 days
Service tier
Standard service
Base rateDefault tier for chat and memory. Requests queue under load and bill at the standard product rate.
Service tier
Priority service
2x only when deliveredAvailable on chat and memory. If unavailable, the request falls back to standard billing and standard handling.
Pick the product
Use Hosted Chat for stateless traffic. Use Hosted Memory for persistent workflows.
Pick the service tier
Standard is the base rate. Priority is 2x only when it is actually delivered.
Pick memory retention
Memory uses 7-day work retention by default. Project retention keeps 30 days and adds $0.50 per request.
Products
Compare the hosted products directly: chat, memory with 7-day retention, and memory with 30-day retention, each available at standard or priority service.
| Decision point | Hosted Chat Standard | Hosted Chat Priority | Hosted Memory 7 days / standard | Hosted Memory 7 days / priority | Hosted Memory 30 days / standard | Hosted Memory 30 days / priority |
|---|---|---|---|---|---|---|
| Best for | One-off requests | Latency-sensitive traffic | Persistent agents | Urgent persistent agents | Longer-running projects | Urgent longer-running projects |
| Input price | $5 / 1M | $10 / 1M* | $10 / 1M | $20 / 1M* | $10 / 1M | $20 / 1M* |
| Output price | $30 / 1M | $60 / 1M* | $60 / 1M | $120 / 1M* | $60 / 1M | $120 / 1M* |
| Service tier | Standard | Priority | Standard | Priority | Standard | Priority |
| Memory retention | None | None | 7 days | 7 days | 30 days | 30 days |
| Extra memory fee | None | None | None | None | +$0.50 / request | +$0.50 / request |
| Queue treatment | Standard queue | Priority queue when available | Standard queue | Priority queue when available | Standard queue | Priority queue when available |
* Priority pricing is 2x only when priority is actually delivered. If
priority is unavailable, billing falls back to the standard product for
that same request. Hosted Memory uses 7-day work retention
by default; 30-day project retention adds $0.50
per request.
Priority billing stays explicit
Request standard or priority service. You only pay the priority rate when priority is actually delivered.