OpenAI-compatible workflow
Single base URL, single bearer token, and SDK-friendly request shapes for the broadest tool support.
- Chat Completions
- Embeddings
- Vision inputs
- Future Responses support
Model Access
Five clear models, one key, and a disclosed quota rate for each.
Choose Sol for frontier work, Terra for balanced daily work, Luna for high-volume light tasks, or keep GPT-5.5 and GPT-5.4 for compatibility. Every subscription tier includes all five.
Model-facing promise
The user should feel they are buying access to capability, not backend complexity.
InputOne key
One account
One account
Unified
Model Route
Model Route
GPT-5.6 Sol
GPT-5.5
GPT-5.4
GPT-5.6 Terra
GPT-5.6 Luna
Coding-first routing
Claude Code, Codex, Cursor, Cline, and similar tools should route through one product story rather than separate provider accounts.
- Claude Code base URL notes
- Codex CLI config
- Cursor custom provider
- Cline provider setup
Provider complexity hidden behind the product
Customers buy capability and billing clarity. Channel selection, routing, and fallback stay behind the business layer.
- Priority routing
- Fallback channels
- Quota-aware grouping
- Future regional tuning
Reference model rates
Reference rates only, shown as USD per million tokens to help you compare models. On a subscription plan you are not billed per token — usage counts against your daily / weekly quota.
| Model | Model ID | Quota rate | Input / 1M | Cached / 1M | Output / 1M |
|---|---|---|---|---|---|
| GPT-5.6 Sol | gpt-5.6-sol | 1.00× | $5.00 | $0.50 | $30.00 |
| GPT-5.5 | gpt-5.5 | 1.00× | $5.00 | $0.50 | $30.00 |
| GPT-5.4 | gpt-5.4 | 0.50× | $2.50 | $0.25 | $15.00 |
| GPT-5.6 Terra | gpt-5.6-terra | 0.40× | $2.00 | $0.20 | $12.00 |
| GPT-5.6 Luna | gpt-5.6-luna | 0.04× | $0.20 | $0.02 | $1.20 |
Provider routing policy
OpenMagic should explain what the user buys without exposing every operational detail of upstream accounts and exit nodes.
Official-first supply
The operating target is official accounts, enterprise purchasing, or compliant upstream channels. Gray account pools are not the product direction.
Node-bound accounts
Provider accounts can be bound to specific exit nodes so traffic can keep a consistent region and IP profile.
Fallback without user changes
Users keep one Base URL. Operations can change channels, model mappings, and node routes behind the gateway.
Expected route patterns
The front-end story has to match the actual request shapes developers paste into tools.
Standard SDKs
Most SDK and client traffic should point to https://api.openmagic.ai/v1 so migration feels obvious.
Anthropic-style exceptions
Some clients will need a root host or alternate path shape. Those exceptions belong in docs, not on the homepage.
Visible model naming
Aliases can simplify the catalog, but they should not hide what customers are actually buying or testing.
Transparent limits
Rate limits, plan caps, and fallback behavior should be explained before users hit avoidable errors.