<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: leandro leite</title>
    <description>The latest articles on DEV Community by leandro leite (@leandro_leite_e6a19fcb1ce).</description>
    <link>https://dev.to/leandro_leite_e6a19fcb1ce</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4240169%2F529b40a8-8565-4842-8694-dad7e1b06e4b.jpg</url>
      <title>DEV Community: leandro leite</title>
      <link>https://dev.to/leandro_leite_e6a19fcb1ce</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/leandro_leite_e6a19fcb1ce"/>
    <language>en</language>
    <item>
      <title>I put a GPU box behind HTTP 402. Two days of real numbers.</title>
      <dc:creator>leandro leite</dc:creator>
      <pubDate>Thu, 12 Nov 2026 03:35:18 +0000</pubDate>
      <link>https://dev.to/leandro_leite_e6a19fcb1ce/i-put-a-gpu-box-behind-http-402-two-days-of-real-numbers-5089</link>
      <guid>https://dev.to/leandro_leite_e6a19fcb1ce/i-put-a-gpu-box-behind-http-402-two-days-of-real-numbers-5089</guid>
      <description>&lt;p&gt;I have a machine at home running open-weights models. This week I put it behind &lt;a href="https://www.x402.org" rel="noopener noreferrer"&gt;x402&lt;/a&gt;, the protocol that turns HTTP status 402 into an actual payment step: the client calls an endpoint, gets a price back, signs a USDC transfer, retries, and gets the answer. No account, no API key, no invoice.&lt;/p&gt;

&lt;p&gt;I wanted to know one thing: &lt;strong&gt;do agents actually pay for things, or is that just a slide?&lt;/strong&gt; So I shipped it on mainnet and counted.&lt;/p&gt;

&lt;h2&gt;
  
  
  What is for sale
&lt;/h2&gt;

&lt;p&gt;Four endpoints, paid per call in USDC on Base:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Endpoint&lt;/th&gt;
&lt;th&gt;Price&lt;/th&gt;
&lt;th&gt;What you get&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;POST /v1/chat&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;$0.005&lt;/td&gt;
&lt;td&gt;Qwen3 14B completion, up to 512 tokens&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;POST /v1/chat-mini&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;$0.002&lt;/td&gt;
&lt;td&gt;Up to 128 tokens, for classify / extract / yes-no&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;POST /v1/embed&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;$0.002&lt;/td&gt;
&lt;td&gt;BGE-M3 embeddings, 1024-dim, up to 16 texts per call&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;POST /v1/vision&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;$0.01&lt;/td&gt;
&lt;td&gt;Image description and OCR, base64 PNG/JPEG&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Base URL: &lt;code&gt;https://x402.llnuvemlocal.com.br&lt;/code&gt;. There is an &lt;a href="https://x402.llnuvemlocal.com.br/llms.txt" rel="noopener noreferrer"&gt;&lt;code&gt;llms.txt&lt;/code&gt;&lt;/a&gt; and an &lt;a href="https://x402.llnuvemlocal.com.br/openapi.json" rel="noopener noreferrer"&gt;&lt;code&gt;openapi.json&lt;/code&gt;&lt;/a&gt; if you would rather let your agent read it.&lt;/p&gt;

&lt;h2&gt;
  
  
  The numbers, unflattering
&lt;/h2&gt;

&lt;p&gt;After roughly two days on mainnet:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;About 2,800&lt;/strong&gt; requests that got a 402 quote and walked away.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;15&lt;/strong&gt; paid calls. Eight of those were me, testing.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;7&lt;/strong&gt; paid calls from strangers, from &lt;strong&gt;6&lt;/strong&gt; different wallets.&lt;/li&gt;
&lt;li&gt;Six of the seven were embeddings. One was chat. Nobody bought vision.&lt;/li&gt;
&lt;li&gt;Revenue from strangers: &lt;strong&gt;$0.017&lt;/strong&gt;.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;So yes, agents pay. They pay once, two tenths of a cent, and so far only one came back.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I learned
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Crawlers ask for discovery files before they pay.&lt;/strong&gt; In the first hours, bots were requesting &lt;code&gt;/.well-known/x402&lt;/code&gt;, &lt;code&gt;/openapi.json&lt;/code&gt;, &lt;code&gt;/llms.txt&lt;/code&gt; and &lt;code&gt;/agents.json&lt;/code&gt; and getting 404. They also probe paid routes with &lt;code&gt;GET&lt;/code&gt; first. I now serve all of those for free, and &lt;code&gt;GET&lt;/code&gt; on a paid route returns a description of how to call it instead of a 405.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Your endpoint is invisible until the first sale.&lt;/strong&gt; The Coinbase facilitator's catalog indexes a route only after it settles a payment. Each route needs its own first purchase, so I bought from myself once per route.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Embeddings are what machines buy.&lt;/strong&gt; I expected chat to lead. It did not. An embedding is a commodity with a fixed shape: an agent can evaluate it without reading prose, and $0.002 for sixteen texts is easy to justify.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Never charge for failures.&lt;/strong&gt; With x402 the server settles after the handler succeeds. If the model is busy I return 503 and nobody pays. This matters more than it sounds: an agent that gets charged for an error has no support channel to complain to, it just never returns.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Price has a floor.&lt;/strong&gt; The hosted facilitator charges per settlement after a free monthly allowance, so anything under about $0.002 can end up costing you money.&lt;/p&gt;

&lt;h2&gt;
  
  
  Use it from your own agent: MCP server
&lt;/h2&gt;

&lt;p&gt;Calling x402 by hand means handling the 402, signing EIP-3009 and retrying. To skip that, I wrote a small MCP server. It is one Python file you run with &lt;code&gt;uv&lt;/code&gt;, and it gives Claude, Cursor or any MCP client these tools: &lt;code&gt;chat&lt;/code&gt;, &lt;code&gt;chat_mini&lt;/code&gt;, &lt;code&gt;semantic_rank&lt;/code&gt;, &lt;code&gt;embed&lt;/code&gt;, &lt;code&gt;read_image&lt;/code&gt;, plus free &lt;code&gt;pricing&lt;/code&gt; and &lt;code&gt;wallet&lt;/code&gt;.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;claude mcp add x402-ai -e &lt;span class="nv"&gt;X402_PRIVATE_KEY_FILE&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="nv"&gt;$HOME&lt;/span&gt;/.x402-buyer.key -- uv run &lt;span class="nv"&gt;$HOME&lt;/span&gt;/x402-ai-mcp/server.py
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The one I expect to be most useful is &lt;code&gt;semantic_rank&lt;/code&gt;: give it a query and up to 15 passages, in any mix of languages, and it returns them sorted by similarity for $0.002. It is a cheap way for a big model to delegate "which of these is relevant" instead of reading everything itself.&lt;/p&gt;

&lt;p&gt;It has two guard rails, because handing a wallet to an agent deserves them: a per-session spend cap (default $0.50) and a per-call ceiling (default $0.02) that rejects a quote before anything is signed. Use a dedicated wallet with a few dollars in it.&lt;/p&gt;

&lt;p&gt;Code: &lt;strong&gt;&lt;a href="https://github.com/Leite007/x402-ai-mcp" rel="noopener noreferrer"&gt;https://github.com/Leite007/x402-ai-mcp&lt;/a&gt;&lt;/strong&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Or with plain Python
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight python"&gt;&lt;code&gt;&lt;span class="k"&gt;from&lt;/span&gt; &lt;span class="n"&gt;eth_account&lt;/span&gt; &lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="nc"&gt;Account&lt;/span&gt;
&lt;span class="k"&gt;from&lt;/span&gt; &lt;span class="n"&gt;x402&lt;/span&gt; &lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;x402Client&lt;/span&gt;
&lt;span class="k"&gt;from&lt;/span&gt; &lt;span class="n"&gt;x402&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;http&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;clients&lt;/span&gt; &lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;x402HttpxClient&lt;/span&gt;
&lt;span class="k"&gt;from&lt;/span&gt; &lt;span class="n"&gt;x402&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;mechanisms&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;evm&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;exact&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;register&lt;/span&gt; &lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;register_exact_evm_client&lt;/span&gt;
&lt;span class="k"&gt;from&lt;/span&gt; &lt;span class="n"&gt;x402&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;mechanisms&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;evm&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;signers&lt;/span&gt; &lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="nc"&gt;EthAccountSigner&lt;/span&gt;

&lt;span class="n"&gt;client&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nf"&gt;x402Client&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="nf"&gt;register_exact_evm_client&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;client&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nc"&gt;EthAccountSigner&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nc"&gt;Account&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;from_key&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="no"&gt;KEY&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;networks&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s2"&gt;"eip155:8453"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;

&lt;span class="k"&gt;async&lt;/span&gt; &lt;span class="k"&gt;with&lt;/span&gt; &lt;span class="nf"&gt;x402HttpxClient&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;client&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;timeout&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mi"&gt;300&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;as&lt;/span&gt; &lt;span class="n"&gt;http&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;r&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="k"&gt;await&lt;/span&gt; &lt;span class="n"&gt;http&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;post&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"https://x402.llnuvemlocal.com.br/v1/embed"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                        &lt;span class="n"&gt;json&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="s2"&gt;"input"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"hello world"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"olá mundo"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;print&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;r&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;json&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"dimensions"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;  &lt;span class="c1"&gt;# 1024&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  Is this a business?
&lt;/h2&gt;

&lt;p&gt;Not at these numbers, and I would rather say that than pretend. I set myself a deadline: if there are not at least three wallets that either come back more than ten times or spend a dollar each by November 24, I turn it off. I will post the result either way.&lt;/p&gt;

&lt;p&gt;If you build agents and have a use for cheap embeddings or a throwaway small model, try it and tell me what is missing.&lt;/p&gt;

</description>
      <category>ai</category>
      <category>mcp</category>
      <category>llm</category>
      <category>web3</category>
    </item>
  </channel>
</rss>
