Put VetFlash AI inside your practice software. Clinical chat on our fine-tuned model, consultations turned into clinical notes, and shareable patient chats. Same models and same credits as the app.
$ Base URL
Everything is JSON over HTTPS, authenticated with a bearer token, and priced in the credits you already have.
POST /chat/completionsVetFlash M2 drafts with medicines-database and web lookups, then a senior clinical reviewer corrects it. Stream the reasoning live, or wait for the reviewed answer.
vetflash 10 crvetflow 30 crPOST /scribe/notesSend a transcript or the audio itself. Get back a structured clinical note in Markdown or plain text.
POST /ask/sessionsA patient-aware chat link for owners or colleagues, with an expiry and a hard credit cap.
GET /credits · /meCheck spend before you call, and know exactly whose account a token belongs to.
OAuth 2.0Build for other practices. Vets approve your app once and it works with their account.
Start with a free test key in the API dashboard. Pick only the scopes you need.
Plain HTTPS from any language, or the OpenAI SDK you already use with a different base URL.
Every call lands in your dashboard with status, latency, credits and a request ID.
curl -X POST https://vetflash.io/api/v1/chat/completions \
-H "Authorization: Bearer $VETFLASH_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "vetflash",
"messages": [
{
"role": "user",
"content": "Maintenance fluid rate for a 12 kg dog?"
}
],
"max_credits": 20
}'For your own backend and scripts.
vf_live_ and free vf_test_ keysAuthorization: Bearer vf_live_...For apps other vets connect to.
Authorization: Bearer vfat_...Your app sends the vet to our consent screen with a PKCE challenge.
They see exactly what your app can do, and that it may spend credits.
Your server swaps the one-time code and verifier for tokens.
Act on their behalf. Tokens refresh and rotate; vets can disconnect anytime.
profile:readcredits:readchat:writescribe:writeask:readask:writeThe API spends the credits on your VetFlash plan at exactly the in-app rate. No separate contract, no metered bill at the end of the month.
max_credits guard. A request that would cost more is rejected before it runs./chat/completions · vetflash/chat/completions · vetflow/scribe/notes/audio/transcriptions/ask/sessions/creditsKeys, secrets, codes and tokens are stored only as SHA-256 hashes. We can't read them back, and neither can anyone else.
S256 is mandatory, codes are single-use, and a replayed refresh token revokes the whole grant.
Keys and tokens carry only the scopes you grant. A missing scope is a clean 403, never a surprise.
60 requests a minute per key or connection, with Retry-After on every 429.
Status, latency, credits and a request ID for each call, searchable in your dashboard.
If a model call fails, the credits go straight back. You only pay for answers you get.
Read our Security Annex and DPA.
No. Every VetFlash account can create API keys, and calls spend the credits already on your account. Endpoints tied to a feature, like Scribe, need a plan that includes it.
Exactly like the app. A VetFlash M2 message is 10 message credits, and scribe is 1 scribe credit per started 20 minutes. Set max_credits on a request to cap what it can spend.
Yes. Chat completions and audio transcriptions follow the OpenAI request and response shapes, including streaming. Point the SDK's base URL at the VetFlash API and use a VetFlash key.
Test keys (vf_test_...) exercise auth, scopes, validation and logging but return sample responses and never spend credits. Swap in a live key when you're ready.
That's what OAuth is for. Register an app, send vets through the consent screen, and your app works with their account and their credits. They can disconnect at any time.
Shareable, patient-aware chat links for colleagues or pet owners. You set the expiry and a credit cap, messages bill to you, and you can revoke a link instantly.
Create a free test key and make your first call in under five minutes.