Platform Overview
SDKMAX is an AI Gateway for teams and developers: it doesn't train models — it aggregates the world's leading AI providers (OpenAI, Anthropic Claude, Google Gemini, DeepSeek, Qwen, and more) behind one API endpoint, one auth scheme, and one OpenAI-compatible protocol.
If your team already has code written against the OpenAI API, integrating SDKMAX usually takes two changes:
- Point
base_urlathttps://api.sdkmax.com/v1 - Swap in an SDKMAX-issued API key (starts with
sk-)
Request bodies, response bodies, and the SSE streaming protocol stay OpenAI-standard.
Why an AI Gateway
Integrating multiple AI providers directly tends to produce the same recurring pain:
- Fragmented accounts and keys — a separate signup, key, and bill per provider.
- Protocol drift — different request/response shapes, auth schemes, and error formats mean business code fills up with per-provider branching.
- Availability risk — no fallback when a single provider rate-limits, has an outage, or deprecates a model.
- Opaque cost/usage — spend scattered across providers, hard to see team-wide AI usage in one place.
SDKMAX resolves these at the gateway layer:
| Capability | Description |
|---|---|
| Unified entry point | All model calls go through api.sdkmax.com — your code integrates one interface. |
| Channel aggregation | Upstream accounts/providers are registered as "channels" and routed by group, weight, and health, with automatic failover. |
| Automatic model sync | Channel model lists and pricing changes propagate automatically — you don't hand-maintain a model catalog. |
| OpenAI-compatible | /v1/chat/completions, /v1/embeddings, /v1/images/*, /v1/audio/* and more follow the OpenAI protocol shape. |
| Enterprise-grade management | Teams, multiple API keys, quotas, rate limits, channel health monitoring, usage dashboards. |
Core concepts
- API Key: the credential sent as
Authorization: Bearer sk-xxx, created in the console with configurable quota, expiry, and model/group restrictions. - Model: e.g.
gpt-4o,claude-opus-4-8,gemini-2.5-pro,deepseek-chat— referenced the same way OpenAI'smodelfield works. The authoritative list is whatever/v1/modelsreturns. - Channel: the internal configuration unit that connects the gateway to an upstream account/provider. Not visible to regular callers — managed by platform/enterprise admins, and determines which upstream account actually serves a given model.
- Group: an API key can be restricted to a specific group, limiting it to the channels/models exposed to that group — useful for internal cost or access isolation.
Who it's for
- Product teams integrating multiple LLM providers who don't want to maintain a separate SDK/key per vendor.
- Teams with existing OpenAI-shaped code who want a low-friction switch or multi-cloud fallback.
- Enterprises needing centralized quota and usage control across many internal projects and API keys.
Next steps
- No account yet? Start with Create an Account.
- Already have one? Jump to the Quickstart for a working call in 5 minutes.
- Want the compatibility scope? Read OpenAI-Compatible API.