x402 · BSV

Inference, paid in Bitcoin.

Every model, billed in BSV.
No account to create, no API key to manage.

 

Open the playground →
Claude Code
Claude Code v2.1.273 · headless `claude -p` · ~/bitplan-gateway-demo
Use gateway.bitplan.dev with Muse Spark 1.3 to summarize this repo's README in one sentence. Show the model's answer and what the call cost.
I'll load the gateway skill first, then read the README and make the call.
Skill(gateway)Launching skill: gateway
Now I'll read the README, get a token, and confirm the model id.
Bash(Read README, mint gateway token, look up Muse model)token minted (exit=0, len=226) · meta/muse-spark-1.3 · $1.50/M in, $5.10/M out(ctrl+o to expand)
# bitplan-gateway-demo A tiny demo repository used to try gateway.bitplan.dev from coding agents. It contains a single script that lists the gateway's models, picks one, and asks it a question, paying for the call from a BSV wallet over HTTP 402. ---TOKEN--- exit=0 len=226 03d09a18… ---MODEL--- {"id":"meta/muse-spark-1.3","reasoning":true,"min_max_tokens":1024, "pricing":{"usd_in":1.5,"usd_out":5.1}} … +7 other muse/spark rows
Token minted and the model id is meta/muse-spark-1.3. Making the call now.
Bash(Call Muse Spark 1.3 via gateway to summarize README)200 OK · 74 in / 1,069 out (1,020 reasoning) · x-gateway-charge-sats: 67408(ctrl+o to expand)
--- headers --- x-gateway-balance-sats: 303216 x-gateway-charge-sats: 67408 x-gateway-request-id: fe23692d-5033-4fe8-a498-7919a0874711 --- response --- { "finish": "stop", "content": "This tiny demo repository provides a single script for coding agents to try the Bitplan gateway by listing models, picking one, and asking a question paid from a BSV wallet over HTTP 402.", "usage": {"prompt_tokens":74,"completion_tokens":1069,"reasoning_tokens":1020}, "charge_sats": null }
The model answered. The charge is in the response header but null in the body, so I'll confirm the settled amount from the usage ledger and get the BSV rate for a USD figure.
Bash(Confirm settled charge from usage ledger and fetch BSV rate)settled · 67,408 sats at 15.97 USD/BSV · latency 9095 ms(ctrl+o to expand)
{"id":"fe23692d-5033-4fe8-a498-7919a0874711","status":"settled", "input_tokens":74,"output_tokens":1069,"reasoning_tokens":1020, "cost_usd":"0.008970750000","charge_sats":67408,"hold_sats":141851, "bsv_usd":"15.97000000","latency_ms":9095}
Done. Muse Spark 1.3 summarized the README through gateway.bitplan.dev, and the call settled. > This tiny demo repository provides a single script for coding agents to try the Bitplan gateway by listing models, picking one, and asking a question paid from a BSV wallet over HTTP 402. Settled charge: 67,408 sats ($0.0090 at 15.97 USD/BSV). Hold before settlement: 141,851 sats. Latency: 9.1s. Balance after: 303,216 sats. Almost all of the cost came from the model's hidden reasoning tokens (1,020 of the 1,069 output tokens), not the one-sentence answer.
Claude Code via the skill: one call, 67,408 sats.
Claude Code example session

Quick start

Give your agent this prompt

It discovers the rest on its own.
prompt.md
Use gateway.bitplan.dev for AI inference.

Identity

Your wallet's public key is your account.

Pay as you go

Each call is priced up front, paid from your wallet or credits.

Price

List prices per million tokens, in BSV. Discounts with volume or your own key.

Works in the harness you already use

Claude Code›_CodexCursorOpenCodeClineGrok BuildGemini CLIWindsurfCopilot›_Zed

Why this gateway

What you get

Pay as you go

Pay per call from your wallet, or use credits.

Bring your own key

Use your own provider key and pay less per token.

Honest prices

Prices come straight from each provider's own catalog.

MCP and OpenAI API

One MCP server, or a familiar OpenAI-style API.

Playground

Try any model in the browser right away.

Many providers, one bill

Compare providers and pick the cheapest route.

Setup

Connect your tool

The plugin signs the token and pays each 402 from your wallet.

opencode.json
{  "$schema": "https://opencode.ai/config.json",  "plugin": ["opencode-plugin-bitplan"]}

Then: /connect → BitPlan Gateway → Connect BRC-100 wallet.

Or start from a template

›_
BitPlan Guide
Grok Bot · by Satchmo

A published Grok Bot template: encrypted plans, paid inference, wallet as account.

Add to Grok BotView on x.ai

API

How to use it

Agents use the MCP server; people connect a wallet and use a token.

One URL that authenticates itself, with tools for models, chat, images and credits. It pays its own 402 when a call needs credits.

mcp.json
{  "mcpServers": {    "gateway": { "url": "https://mcp.bitplan.dev" }  }}

Models

Every model, one price list

Prices come straight from each provider's own catalog, in BSV, sats and USD per million tokens. Try any of them in the playground before you write a line of code.

Browse all modelsOpen the playground

Starred by default

anthropic/claude-fable-5.1openai/gpt-6-astragoogle/gemini-3.8-flashspacexai/grok-4.6deepseek/deepseek-v4.1-flashmoonshotai/kimi-k3meta/muse-spark-1.3alibaba/qwen3.8-maxmistral/mistral-large-3zai/glm-5.3minimax/minimax-m3openai/gpt-image-2.5-flare