Pricing. Inference is metered. The open-source side is free.
The gateway charges per token from a prepaid credit balance, or a flat price per request over x402 with no account. The network, OpenClaude, Zero, iCaptcha and the explorer cost nothing. Rates below are read live from the gateway.
Credits. A prepaid balance, drawn down per token.
| Way to pay | Amount | What you get |
|---|---|---|
| Monthly plan | $10 a month | Adds $10 of credits every month, paid by card. Unused credits roll over while you are subscribed; cancel from the billing portal. |
| One-time top-up | You choose | No subscription. Any amount by card, or $5, $10, $25 or $50 in USDC on Base from a wallet. |
| Referrals | $2 each | $2 in credits when someone who signed up with your link makes their first payment, up to $50 a month. |
Credits are prepaid usage, not cash, and are not transferable. A key draws from its account's balance.
Rates per 1M tokens
| Model | Input | Cached input | Output |
|---|---|---|---|
Auto routingauto | Billed at the rate of the model that served it | ||
Qwen 3.8 Maxqwen/qwen3.8-max | $2.40 | $0.30 | $7.20 |
Qwen 3.8 Flashqwen/qwen3.8-flash | $0.18 | $0.0192 | $0.564 |
MiMo V2.6-Proxiaomi/mimo-v2.6-pro | $0.522 | $0.00432 | $1.04 |
DeepSeek V4.1 Flashdeepseek/deepseek-v4.1-flash | $0.0612 | $0.0216 | $0.68 |
Claude Opus 5.5anthropic/claude-opus-5.5 | $4.80 | $0.24 | $24.00 |
Claude Sonnet 5.5anthropic/claude-sonnet-5.5 | $2.40 | $0.12 | $12.00 |
Claude Haiku 5.5anthropic/claude-haiku-5.5 | $0.12 | $0.012 | $0.60 |
GPT-6.1 Solopenai/gpt-6.1-sol | $2.40 | $0.12 | $12.00 |
GPT-6.1 Terraopenai/gpt-6.1-terra | $1.20 | $0.06 | $6.00 |
GPT-6 Lunaopenai/gpt-6-luna | $0.12 | $0.012 | $0.60 |
MiMo V2.6-Flashxiaomi/mimo-v2.6-flash | $0.168 | $0.00336 | $0.336 |
Gemini 3.5 Flash Litegoogle/gemini-3.5-flash-lite | $0.36 | $0.036 | $3.00 |
MiniMax M3minimax/minimax-m3 | $0.36 | $0.072 | $1.44 |
Qwen 3.7 Maxqwen/qwen3.7-max | $1.77 | $0.354 | $5.31 |
Kimi K3.1 Promoonshotai/kimi-k3.1-pro | $2.40 | $0.24 | $12.00 |
GLM 5.3z-ai/glm-5.3 | $1.68 | $0.312 | $5.28 |
Nemotron 3.5 Super VL (free)nvidia/nemotron-3.5-super-vl:freeFree, rate limited; no credits needed | $0 | $0 | $0 |
Nemotron 3.5 Super VLnvidia/nemotron-3.5-super-vl | $0.18 | $0.024 | $0.60 |
Ling 3.1 Flashinclusionai/ling-3.1-flash | $0.09 | $0.018 | $0.27 |
Tencent HY4tencent/hy4 | $0.50 | $0.0252 | $1.50 |
Macaron V1 Tallmindai/macaron-v1-tall | $0.54 | $0.096 | $3.12 |
Macaron V1 Ventimindai/macaron-v1-venti | $1.80 | $0.36 | $5.40 |
Live from the gateway's model list, GET /v1/models, re-read every five minutes. USD per 1M tokens, as billed from a credit balance.
Per request. Over x402, in USDC on Base, with no key and no account.
POST /v1/chat/completionsThe agent calls with no key.402 Payment RequiredThe gateway quotes the price in USDC on Base.POST /v1/chat/completionsThe same call, carrying the signed payment.200 OKThe answer, with the settlement transaction in a header.
| Model | Per request |
|---|---|
Qwen 3.8 Maxqwen/qwen3.8-max | $0.09 USDC |
Qwen 3.8 Flashqwen/qwen3.8-flash | $0.01 USDC |
MiMo V2.6-Proxiaomi/mimo-v2.6-pro | $0.02 USDC |
DeepSeek V4.1 Flashdeepseek/deepseek-v4.1-flash | $0.005 USDC |
Claude Opus 5.5anthropic/claude-opus-5.5 | $0.20 USDC |
Claude Sonnet 5.5anthropic/claude-sonnet-5.5 | $0.10 USDC |
Claude Haiku 5.5anthropic/claude-haiku-5.5 | $0.005 USDC |
GPT-6.1 Solopenai/gpt-6.1-sol | $0.10 USDC |
GPT-6.1 Terraopenai/gpt-6.1-terra | $0.05 USDC |
GPT-6 Lunaopenai/gpt-6-luna | $0.005 USDC |
MiMo V2.6-Flashxiaomi/mimo-v2.6-flash | $0.005 USDC |
Gemini 3.5 Flash Litegoogle/gemini-3.5-flash-lite | $0.015 USDC |
MiniMax M3minimax/minimax-m3 | $0.015 USDC |
Qwen 3.7 Maxqwen/qwen3.7-max | $0.03 USDC |
Kimi K3.1 Promoonshotai/kimi-k3.1-pro | $0.06 USDC |
GLM 5.3z-ai/glm-5.3 | $0.03 USDC |
Nemotron 3.5 Super VLnvidia/nemotron-3.5-super-vl | $0.01 USDC |
Ling 3.1 Flashinclusionai/ling-3.1-flash | $0.005 USDC |
Tencent HY4tencent/hy4 | $0.01 USDC |
Macaron V1 Tallmindai/macaron-v1-tall | $0.02 USDC |
Macaron V1 Ventimindai/macaron-v1-venti | $0.04 USDC |
Live from the gateway's model list, GET /v1/models, re-read every five minutes. One flat price per call, covering up to 4,096 output tokens. Not payable per request: auto, nvidia/nemotron-3.5-super-vl:free.
Works on /v1/chat/completions and /v1/messages. The price for each model is also on GET /v1/models.
Free. What costs nothing, and the limits where there are any.
| The network | Pushing signed code, creating repos and running your own node. The node is open source. |
|---|---|
| OpenClaude | The coding-agent CLI is open source and runs on your machine. You bring the model provider. |
| Zero | The coding agent is open source and runs on your machine. You bring the model provider. |
| iCaptcha | Open source under MIT; run it yourself or call the public instance. |
| Explorer | Browsing the network, and the markdown agents read from it. |
| The free model | Nemotron 3 Ultra (free) answers with a zero balance. It is rate limited and shares an upstream pool, so it can be busy. |
| MiMo V2.6-Pro allowance | With a zero balance, 30 requests per 5-hour window (windows start at 00, 05, 10, 15 and 20 UTC). |
Questions.
- Is the network free?
- Yes. Pushing signed code, creating repos and running your own node cost nothing. The node is open source and self-hostable.
- What does a credit balance buy?
- Gateway inference at the per-token rates above. The cost is deducted per request, and non-streaming responses report it in the
x-gateway-cost-usdheader. - What happens at a zero balance?
- Paid models answer 402 until you top up. The free model and the MiMo allowance keep working.
- Can an agent pay without an account?
- Yes, over x402: the gateway answers with a 402 and a price in USDC on Base, the agent signs the authorization and retries, and the response carries the settlement transaction.
- Refunds?
- Card top-ups follow the refund policy. Payments in USDC, including per-request payments, settle on-chain and are final.