Send opportunities, historical evidence and costs. Receive selected quantities and predicted economics in a result your code can use.
Beta Validation is free and needs no key. Every output on this page is labelled recorded, mock or actual.
POST /v1/validate free · no key
→ {"ok": true, "errors": []}
POST /v1/fit your key
→ {"session_id": "…"}
GET /v1/result/{id} poll · minutes
→ grounding … processing … done
→ {"key": "h10-1519-1055",
"qty": 2, "profit": 9.19}Today's options and past ones — including the options nobody took — plus what happened to the ones that were taken.
How your unit economics work, in plain words: fees, holding costs, write-offs. Approximations are fine.
Zero means pass. Predicted profit per row and for the plan, after a fit that takes minutes — not milliseconds.
The authoritative contract is the live API root, api.hyperc.com/v1/. Field-by-field documentation, error codes and the data floors are in the API documentation; the API brief is the one-page version.
The quickstart validates a versioned sample request, submits it as an input test once you set a key, polls until it finishes and saves the result. It never places an order.
curl -O https://hyperc.com/lp/sample/p34_quickstart.py
python3 p34_quickstart.py # 1 · validate the sample: free, no key
export P34_API_KEY=… # from your environment, never from code
python3 p34_quickstart.py # 2 · input test (mock: true) → p34-out/result-mock.json
python3 p34_quickstart.py --input your-request.json --real # 3 · an actual fit on your records
python3 p34_quickstart.py --resume SESSION_ID # keep polling a running fit| Step | Key | Cost |
|---|---|---|
Validate · POST /validate | None | Free. Nothing is queued or charged. |
Input test · mock: true | Your API key | Free. Returns placeholders, not predictions. |
| Actual fit · your records | Your API key | Pay as you go: $1 per token, typically 20–30 tokens for a first fit on new data. |
API keys come from the API platform (Get API key): pay as you go, no subscription, prices below. Keep the key in an environment variable, never in code, logs or URLs. The sample request (p34-sample-request-v1.json, 300 generated offers from a made-up reseller) is for learning the format: the quickstart refuses --real on it, because an actual fit on generated data would spend compute on numbers that mean nothing. Read the quickstart before you run it.
39 quantity options each — 46,800 choices — plus the market's history.
21 units, $593.14 committed. Predicted profit $59.82.
| Offer | UnitsQty | Unit cost | Cost incl. fulfilment | P34 predicted | Simulator outcome |
|---|---|---|---|---|---|
| h10-1519-1055 | 2 | $26.03 | $57.05 | $9.19 | $7.63 |
| h10-1519-1056 | 2 | $26.68 | $58.36 | $4.26 | $4.49 |
| h10-1519-1060 | 1 | $27.37 | $29.87 | $0.19 | $0.15 |
| h10-1519-1061 | 2 | $25.97 | $56.95 | $6.65 | $3.37 |
| h10-1519-1064 | 2 | $26.17 | $57.34 | $6.08 | $6.32 |
| h10-1519-1065 | 2 | $25.28 | $55.55 | $6.61 | $6.78 |
| h10-1519-1071 | 2 | $25.35 | $55.69 | $7.08 | $7.26 |
| h10-1519-1119 | 2 | $25.92 | $56.85 | $4.42 | $7.30 |
| h10-1519-1189 | 3 | $25.50 | $84.00 | $7.16 | $10.15 |
| h10-1519-1194 | 3 | $24.66 | $81.48 | $8.18 | $13.52 |
| 1,190 other offers | 0 | — | — | — | — |
| Plan total | 21 | $593.14 | $59.82 | $66.98 |
Recorded run, not generated now. P34 model rc012.2 on t5market, a game-item market whose outcomes come from a deterministic simulator — so the true result of every offer can be computed, including the ones nobody bought. Market h10, tick 1519, 26 September 2026 (session 9e6c8df2). “Simulator outcome” is the full-horizon counterfactual for exactly these quantities: not an executed trade, and not live-market evidence. It excludes the venue's monthly participation fee. One run, shown because it was verified end to end — not a benchmark. Rows are rounded to the cent; totals use unrounded values. Download the plan (CSV) · Benchmarks
{"status": "done", "session_id": "9e6c8df2…", "model": "rc012.2",
"menu": [{"key": "h10-1519-1055", "menu": 0, "T": 0, "qty": 2.0, "profit": 9.192633},
… 10 rows with qty > 0; any key absent means do not trade],
"n_selected": 10, "predicted_profit_sum": 59.817638, "confidence_sweep": [ … ]}| Label | What it is |
|---|---|
| Recorded | The result on this page: real model output from 26 September 2026, stored and shown again. Not generated now. |
| Mock | An input test. Placeholders in the real shape, marked "mock": true. Never a prediction. |
| Actual | A new fit on your data. Minutes, not milliseconds; free on free-tier markets, tokens elsewhere. |
P34 is a component your application calls at the moment it has to choose between commercial actions — not a framework to build around.
| Piece | What it does |
|---|---|
| P34 | The hosted model. Evaluates a set of opportunities and returns quantities and predicted economics. |
| Your agent or program | Collects information, prepares the inputs, calls P34, shows you the result, carries out steps you authorize. |
| Your agent harness | Manages the agent's tools and workflow. P34 is something it calls — not a replacement for it. |
| HyperC workspace | A managed machine where the agent's files, tools and work live, running 24/7. Optional until you want it kept running. |
| P34 skills | Instructions that help an agent prepare a valid request and read the answer. Not the model itself. |
Add a P34 decision step to this project. Read https://api.hyperc.com/v1/ and https://github.com/hyperc-ai/P34-API-DOCS first. Map our data into Menus (current options plus history, including the options we did not take) and Sales (what happened to the ones we took), and write a business description of our unit economics. Validate with POST /validate (no key) until there are no errors. Read the key from the P34_API_KEY environment variable; never write it into code, logs or URLs. Submit with mock: true first, poll GET /result/{session_id} until done or failed, and save the response. Show the qty > 0 rows as a plan for human review. Do not place orders.
It starts with validation and an input test, keeps the key in the environment, and stops at a plan for review.
The P34 API on its own. Registration is free and there is no subscription: buy tokens from $15 and actual fits spend them — typically 20–30 for a first fit on new data. Everything before a fit is free.
POST /validate with no key, and input tests with mock: trueUsage-based software service, paid in tokens. No profit is promised.
| What | Cost |
|---|---|
| Validate a request | Free |
| Input test | Free |
| Grounding | typically 17–20 tokens |
| Grounding, repeated | replay metered as compute |
| Model fit | typically 3–10 tokens |
| A failure on our side | Not charged |
None to validate. An API key runs input tests free and actual fits in tokens, pay as you go — no subscription and no workspace needed.
Menus — today's options and past ones, including the options nobody took — and Sales, what happened to the ones that were taken. Plus a business description: how your unit economics work, in plain words. POST /validate says what is wrong before you spend anything.
mock: true checks your inputs and returns placeholders in the real shape, free on every tier — never use them as predictions. An actual fit grounds your history, runs the model and takes minutes.
It is a REST API: JSON in, JSON out. The quickstart is one Python file that uses only the standard library; the documentation repository has a fuller client, tests and the data format.
A quantity for each opportunity — zero means pass — and predicted profit, calibrated as a total for the plan. Predicted is a forecast, not a result. The answer does not include a reason for each pass.
No. A run takes minutes. It is built for purchasing-style decisions, not high-frequency trading.
P34 is built to choose and size across a whole set of opportunities, using business history and costs, and its plans can be checked against what happened next. History usually records outcomes for only some of the choices; P34 is designed around that gap. Your agent still does the research, communication and workflow.
Validation and input tests are free. Actual fits spend tokens at US$1 each — typically 20–30 for a first fit on new data, less when the business description and data schema repeat. Tokens you buy never expire. The full price list is on the API platform page.
Start with a shadow test on your own data, then a capped pilot, then scale with controls. The enterprise track is at hyperc.com/enterprise.
Not ready for a key yet? Join the free AI-economy fan community and get notified about price drops and promotions.
Follow the self-driving business movement at no cost: market experiments as we run them, member findings, P34 and agent-tool releases, research on the emerging AI economy — and first access when the P34 Community opens.
Free forever · no spam · unsubscribe anytime · Privacy Notice
Usage-based API access, paid in tokens; no subscription. Built to pursue business outcomes; no profit is promised.