AI models & inference

Call OpenAI's gpt-5.4 via a single pay-per-call x402 endpoint — no OpenAI account or API key needed, pay $0.25 per request in USDC. Standard OpenAI chat.completions request/response shape, capped input and output.

Call OpenAI's gpt-5.6-terra via a single pay-per-call x402 endpoint — no OpenAI account or API key needed, pay $0.25 per request in USDC. Standard OpenAI chat.completions request/response shape, capped input and output.

Call OpenAI's gpt-5.6-sol via a single pay-per-call x402 endpoint — no OpenAI account or API key needed, pay $0.65 per request in USDC. Standard OpenAI chat.completions request/response shape, capped input and output.

Generate a premium image from a text prompt using GPT Image 2 (medium quality, 1024x1024). Flagship model with best detail and coherence. No API key needed; pay per call via x402. Returns base64 PNG. Prompt capped at 4000 chars.

OpenRouter — one endpoint for 100+ LLMs, keyless pay-per-call (price varies by model + tokens). Params: `model` (vendor/model id like openai/gpt-4o), `messages`.

OpenAI-compatible chat completions over x402, answered by Claude Sonnet 4.6 — send a messages array and get an assistant reply for Q&A, reasoning, agent planning, coding help, summarization, extraction, and classification. Body is OpenAI chat.completions-shaped ({model?, messages:[{role,content}], max_tokens?, temperature?}); response is a chat.completion object. Flat $0.06; inputs capped ~6KB, max_tokens 1024; no streaming in v1.

Generate agent-ready image assets (icons, logos, social graphics, thumbnails, banners) with gpt-image-1. Claude refines the prompt, screens content, adds alt text and a score. Returns image_url as a base64 PNG data URI. Flat $0.25; n=1; ratios 1:1/16:9/9:16.

Language detection for text you send: one of 12 European languages plus Japanese, Chinese, Korean, Arabic and Russian by script, with a confidence score. Route text to the right model or translator. $0.01 per call.

llms.txt reader: checks whether a site publishes llms.txt or llms-full.txt (the emerging standard for telling AI agents what a site offers), returns the content (up to 20 KB) and its headings and links. Discover agent-friendly documentation instantly. $0.01 per domain.

Page language detection: the declared html lang and content-language, the language detected from the text itself (12 European languages plus CJK/Arabic scripts), agreement between the two, and the hreflang alternates count. Route content to the right model or translator. $0.01 per page.

Page to RAG chunks: extracts the main content and splits it into overlapping chunks of about 400 words (roughly 500 tokens) on paragraph and sentence boundaries, each with its index, word count and the nearest heading. Ready for embedding pipelines. $0.01 per page.

Agent discovery card fetch: looks for /.well-known/agent.json and /.well-known/agent-card.json (Google A2A Agent Card) plus /llms.txt and /.well-known/ai-plugin.json, and summarises whatever is found: name, description, url, version, capabilities (streaming, push), skills (id, name, description, tags), auth schemes, input/output modes, and provider. Lets agents learn what another agent offers. $0.01 per domain.

Is a specific bot allowed to fetch a URL? Fetches the site's robots.txt and evaluates the path for the named user-agent (GPTBot, ClaudeBot, Googlebot, Bingbot, CCBot, PerplexityBot or any token), returning allowed/disallowed, the matching rule and group, crawl-delay, and whether the bot has a dedicated group or falls back to *. Pre-flight for compliant scraping agents. $0.01 per check.

MCP server probe (streamable HTTP): runs initialize and tools/list against a Model Context Protocol server and returns protocol version, server name/version, capabilities, instructions, tool count and the tools (name, description, input properties), whether a session id is issued, whether it answers with JSON or SSE, and if the server requires OAuth (401 + resource metadata). Discover what an agent can call before wiring it in. $0.01 per server.

$0.0002/MB. Infer a JSON Schema or SQL DDL straight from your data — skip hand-writing schemas. 1¢ x402 min; remainder auto-credits — relaystation.ai/penny

Get raw book ticker data at microsecond granularity

Get OHLCV candlestick data for a Binance trading pair

AI chat completions API - pay per request with USDC on Base

Google's native generateContent wire on the premium tier. Send Gemini's own request shape (contents, systemInstruction, generationConfig, function declarations) and get Gemini's own response shape (candidates, usageMetadata) back. Point a Google GenAI SDK at this gateway and pay per request with USDC, no account. Also answers the Google-shaped path /v1beta/models/<model>:generateContent.

Lee 2012 EHMRG 7-day ED model (Table 3 integer score): 2*age + 60 if EMS transport - min(SBP,160) + clamp(HR,80-120) - 2*min(SpO2,92) + 20*creatinine_mg_dl + K (<=3.9: +5; 4.0-4.5: 0; >=4.6: +30) + 60 if troponin>ULN + 45 if active cancer + 60 if home metolazone + 12. Groups 1/2/3/4/5a/5b. Dialysis-dependent excluded. Not EHMRG30-ST, not ADHERE CART, not a discharge or admission order.

Khorana 2008 Blood predictive model for chemotherapy-associated thrombosis: very_high site (stomach or pancreas) 2, high (lung, lymphoma, gynecologic, bladder, testicular) 1, other 0; platelets >=350 1; hemoglobin <10 g/dL or ESA 1 (once); WBC >11 x10^9/L 1; BMI>=35 caller flag 1. Max 6. 0 low; 1-2 intermediate; >=3 high. Not a diagnosis. Not Padua or Caprini. Caller assigns cancer site; magent does not compute BMI.

Fonarow 2005 ADHERE CART bedside tree from admission BUN, systolic BP, and serum creatinine. high: BUN>=43, SBP<115, Cr>=2.75; intermediate_1: BUN>=43, SBP<115, Cr<2.75; intermediate_2: BUN>=43, SBP>=115; intermediate_3: BUN<43, SBP<115; low: BUN<43, SBP>=115. Creatinine consulted only when BUN>=43 and SBP<115. CART tree only, not the logistic model or GWTG-HF. Not a heart-failure diagnosis or treatment/ICU order.

Change watch for text chunker: runs the check and compares it with the result from your previous call for the same url (kept 90 days), returning the current answer, changed true/false, exactly which fields changed with before/after values, and when it was last seen. Call it on a schedule to be told when page to rag chunks changes. $0.01 per check.

Change watch for llms txt fetch: runs the check and compares it with the result from your previous call for the same domain (kept 90 days), returning the current answer, changed true/false, exactly which fields changed with before/after values, and when it was last seen. Call it on a schedule to be told when llms changes. $0.01 per check.

Bulk llms txt fetch: up to 20 domains in one call, processed concurrently, results returned in input order with a per-item error field and a count of failures. Same answer per item as the single /site/llms-txt endpoint (llms). Batch enrichment for agents that hold a list. $0.01 per batch.

AI chat completions API - pay per request with USDC on Base

Introductory pricing. NeuroDynamic speech-to-text transcription: convert a voice note or short audio recording into a JSON transcript. Up to 60 seconds and 12 MiB; WAV, MP3, FLAC or Ogg. No account needed. Read https://api.neurodynamic.tech/docs for inputs, privacy and payment recovery. 0.01 USDC on Base.

Business Frameworks — Manage the cash cycle, applied to the caller's own numbers. Send the three day-counts and the cost shares; get the cycle, the cents tied up per sales dollar, and the lever to pull first. Deterministic (the node's rule, no model); returns the worked decision plus the node text and its public citations. Schema at GET /apply/cash-flow/decide/cash-cycle.

Business Frameworks — Build for transfer or run for cash, applied to the caller's own numbers. Send whether the position is durable and the founder-free and repeating revenue shares; get build-for-transfer or run-for-cash, the matching actions, and the quarterly check. Deterministic (the node's rule, no model); returns the worked decision plus the node text and its public citations. Schema at GET /apply/value/decide/build-or-run-for-cash.

Business Frameworks — Reinvest or distribute, applied to the caller's own numbers. Send r, the cash available, the runway target and the candidate uses with their returns; get them ranked, the ones to fund, where to stop, and what to distribute. Deterministic (the node's rule, no model); returns the worked decision plus the node text and its public citations. Schema at GET /apply/value/decide/reinvest-or-distribute.