Aimyria API · pricing and routing
Routes from 0.03x. Know what you pay for.
Aimyria brings compatible model routes into one API endpoint. The catalog currently shows a GPT-Plus promotional group from 0.03x; the marked promotion may change, while the available model, protocol and final charge still depend on the route you select.
Which low-multiplier routes are shown?
The following examples come from public catalog version 12, observed on 2026-09-24. A group may support only certain models or account conditions; the catalog can change.
| Catalog group | Shown multiplier | What to check |
|---|---|---|
| GPT-Plus promotional group | 0.03x | Marked as intermittent; do not assume permanent availability. |
| GPT-Plus regular group | 0.15x | Confirm the current model and group. |
| GPT-Pro group | 0.25x | A separate route from GPT-Plus. |
| Claude Kiro group | 0.21x | Confirm the Anthropic-compatible request format. |
| Domestic-model Lite group | 0.40x | Copy the exact model ID from the live catalog. |
| Mainstream domestic-model group | 0.50x | Copy the exact model ID from the live catalog. |
How is the final cost determined?
The group multiplier is one billing factor. The listed model input/output prices, selected group, actual token usage, account terms and applicable promotions also matter. A multiplier alone does not establish a final cash price or a universal percentage of an official provider's bill. Before paying, compare the live model catalog and the billing display in your account.
Does a lower multiplier mean a lower-quality model?
The multiplier is a pricing setting, not a model-capability setting. The catalog labels one GPT-Pro group “stable · no intelligence reduction”; that is a catalog label, not an independent quality benchmark. Verify the exact model ID, API format, context and tool features, then compare responses using representative prompts. Model revisions, sampling parameters and upstream availability can also change results.
Is every route fast?
Latency depends on the model, request length, network and upstream load. The public catalog does not include a comparable all-route latency benchmark. Measure time to first token and total response time with the same prompts and settings before moving production traffic. The public Python and cURL examples give you a small starting request.
Start with one small request
After checking the supported regions, choose a current model and route, create your own private key, and send a short request. Review the model ID, response and usage record before scaling up.