Understand the API contract
Use the stable https://oppermind.com/api/v1 base, public Lato model identifiers, and modality-specific endpoints.
✦ Course · Beginner
Oppermind Developer API quickstart
Create a scoped API key, make your first Lato text request, read usage metadata, and handle failures safely from a backend.
Start courseYOUR OUTCOME
You will make a secure server-side API call and turn the response into a reliable application path.
Keep keys on the server Use the versioned v1 contract Build for retries, limits, and stable errors
5 min
Create a least-privilege key
Create an opmd_sk_ key in the Developer console, choose only the permissions needed, copy it once, and store it in a server secret.
Send your first text request
POST messages to /api/v1/messages with oppermind-lato-1, a bounded max_tokens value, and stream set to false or omitted.
Read the response and usage
Extract text content, capture the request ID, inspect normalized model metadata, and record input and output token usage.
Generate an image
POST a prompt to /api/v1/images and choose a URL or base64 response according to your storage and delivery path.
Generate a video
POST to /api/v1/videos with a generous client timeout; generation is synchronous and returns the finished authenticated proxy URL.
Handle the unhappy paths
Use idempotency keys, RateLimit headers, stable error codes, safe retries, billing checks, and structured logging.
What each lesson covers
The course, lesson by lesson.
Understand the API contract
Use the stable https://oppermind.com/api/v1 base, public Lato model identifiers, and modality-specific endpoints.
Point your client at https://oppermind.com/api/v1 and note the three endpoints you will use: POST /messages for text, POST /images for images and POST /videos for video. Send the model id oppermind-lato-1 on every text request.
Create a least-privilege key
Create an opmd_sk_ key in the Developer console, choose only the permissions needed, copy it once, and store it in a server secret.
In the Developer console, create a key named support-summariser-prod, choose the text only permission, copy the opmd_sk_… value once and store it in your server's secret manager. Never paste it into browser or mobile code.
Send your first text request
POST messages to /api/v1/messages with oppermind-lato-1, a bounded max_tokens value, and stream set to false or omitted.
curl https://oppermind.com/api/v1/messages \ -H 'Authorization: Bearer opmd_sk_…' \ -H 'Content-Type: application/json' \ -d '{"model":"oppermind-lato-1","max_tokens":200,"messages":[{"role":"user","content":"Summarise our 30-day returns policy in two sentences for a product page."}]}'
Read the response and usage
Extract text content, capture the request ID, inspect normalized model metadata, and record input and output token usage.
From that response, log in your server: content[0].text, the id, the model value, usage.input_tokens and usage.output_tokens, plus the X-Request-ID and X-API-Version response headers.
Generate an image
POST a prompt to /api/v1/images and choose a URL or base64 response according to your storage and delivery path.
POST https://oppermind.com/api/v1/images with the body {"prompt":"Flat-lay photo of a linen tote bag on a timber bench in soft morning light","n":1,"quality":"hd","aspect_ratio":"1:1","response_format":"url"} Use b64_json instead when you store the bytes yourself rather than fetching a URL.
Generate a video
POST to /api/v1/videos with a generous client timeout; generation is synchronous and returns the finished authenticated proxy URL.
POST https://oppermind.com/api/v1/videos with a client timeout of several minutes and the body {"prompt":"Slow aerial pan over rows of solar panels at sunrise","duration_seconds":10,"resolution":"1080p","fps":24}
Handle the unhappy paths
Use idempotency keys, RateLimit headers, stable error codes, safe retries, billing checks, and structured logging.
Send the same POST /messages twice with the header Idempotency-Key: order-4471-summary-v1, and in your handler branch on error.code: back off with jitter on OPMD_RATE_001 after reading the RateLimit headers, stop and alert on OPMD_BILLING_001, never retry a 400 unchanged, and log X-Request-ID on every response.
Keep going
Make the learning stick
Start with one real piece of work.
Open Oppermind beside the lesson, apply each step, and leave with something you can use.