Ship more than one AI feature
Add copy generation and product images through the same platform and credentials. Use the endpoint for each modality from your existing backend.
See the two API requests →Explore generated scenes in Creator, then use the developer examples below to evaluate text and image features for your application. The examples show individual model outputs, not an automated campaign workflow.
Example workflow
The brief · Launch a sculptural cobalt-blue desk lamp with an English headline and a widescreen product image.

Nano Banana 2 · 1376 × 768 PNG
Launch headline · Claude Opus 5
Luma: Cobalt-Blue Aluminum, Brilliantly Focused Light
Two server-side requests turn the brief into a headline and a product image.
See both requests →80 text tokens · 1 generated image
See usage by request →Add copy generation and product images through the same platform and credentials. Use the endpoint for each modality from your existing backend.
See the two API requests →Try a representative brief before wiring the feature into your app. Compare the outputs against what your users need, then choose the model for each step.
Try the Playground →Use API response metrics and Token360 billing records to understand spend. Tie requests to your app’s customer and feature IDs to calculate unit costs.
See measured usage →Set TOKEN360_API_KEY in your terminal. In production, send both requests from your backend and keep the key there.
curl --fail-with-body https://api.token360.ai/v1/chat/completions \
-H "Authorization: Bearer $TOKEN360_API_KEY" \
-H "Content-Type: application/json" \
--data '{
"model": "claude-opus-5",
"messages": [
{
"role": "user",
"content": "Write one short English launch headline for a fictional desk lamp named Luma, made from cobalt-blue anodized aluminum. Return only the headline."
}
]
}'Returned headline: Luma: Cobalt-Blue Aluminum, Brilliantly Focused Light
curl --fail-with-body https://api.token360.ai/v1/images/generations \
-H "Authorization: Bearer $TOKEN360_API_KEY" \
-H "Content-Type: application/json" \
--data '{
"model": "nano-banana-2",
"n": 1,
"aspect_ratio": "16:9",
"prompt": "Editorial launch photograph of Luma, a fictional sculptural cobalt-blue anodized aluminum desk lamp on a pale limestone plinth. Warm light pools under its conical shade, late-afternoon shadows across an ivory wall, strong blue silhouette, wide 16:9 composition. No text or logos."
}'Returned one product image. The preview above displays the downloaded 1376 × 768 PNG.
These figures come from the Luma API responses. For campaign cost, record a campaign ID alongside each request in your app and reconcile the two charges against Token360 billing records.

Customer
Can I change my delivery address?
Reply draft
Send us your order number and we’ll check whether the address can still be updated.
Give support teams an editable reply based on the selected conversation.
Extract invoice fields and flag ambiguous values for review.
When a customer submits a campaign brief, create a campaign record in your backend. Save the headline and image against that record so the customer can review them together and regenerate either one separately.
Check that the headline and image are usable. Store them under the campaign ID.
Keep successful outputs. After a timeout, check the original request before retrying.
Let customers edit and approve the copy and image before exporting.
For Luma, the two successful responses reported 80 text tokens and one generated image. Use billed charges to calculate campaign cost. For reliability, track how often both steps finish and how long customers wait for the complete draft.
Self-serve
Get an API key, then run the two Luma requests from your terminal.
POST /v1/chat/completionsLuma: Cobalt-Blue Aluminum, Brilliantly Focused Light
For teams
Discuss capacity, access controls, and billing with the Token360 team.
Use /v1/chat/completions for the headline and /v1/images/generations for the image.
Store it in your backend environment or secret manager. Your browser calls your own backend.
Associate each request with a customer ID in your app, then reconcile its billed charge before aggregating by customer or feature.
Save the successful output and show the status of the remaining step. Let the customer retry just that step when it is safe to do so.
Run a fixed set of representative briefs through both models. Review output quality, latency, and billed cost before changing the model for customers.