---
title: "Submit a prediction"
method: POST
path: "/predict"
tags: ["Predict"]
---

# Submit a prediction

`POST /predict`

Reserves credits atomically and enqueues preprocessing. Returns 202 + `jobId`. On insufficient balance returns 402 with rollback. On per-org concurrency cap returns 429 with `Retry-After`.

## Request body

- object
  - `uploadId` string, uuid, required
  - `markers` string[], required
  - `model` 'v0.4' | 'v0.5' — Optional explicit Lattice version. `v0.4` is the 192-marker original; `v0.5` is the retrained 192-marker sibling and the current default. Both share GenePT embeddings — the marker vocabulary is identical between them. Omit to let the platform pick (currently `v0.5`). Legacy strings `"v10-fullpanel"` and `"v10-fullpanel-v2"` are still accepted on input and rewritten server-side to `"v0.4"` / `"v0.5"` respectively (with a `Warning: 299` response header); the legacy alias path is undocumented and will be removed on 2026-12-01. The `"v10"` alias resolved to the now-sunset v0.3 and is rejected with 400 `unknown_model`.

## Response `202`

Job accepted and credits reserved

- Submission
  - `jobId` string, uuid, required
  - `reservedCredits` integer, required
  - `status` 'queued', required

## Other responses

- `402` — Insufficient credits
- `409` — Upload not ready
- `429` — Per-org concurrent job cap exceeded

---

[API](https://skmtc.dev/strandai/apis/strand-ai-platform-api.md) · [All operations](https://skmtc.dev/strandai/apis/strand-ai-platform-api/llms.txt) · [OpenAPI document](https://skmtc-service-production.skmtc.workers.dev/v1/apis/strandai/strand-ai-platform-api/revisions/811cda19da65/schema)