LLM API for GPT, Claude and Gemini with one key

The Omni API gives you OpenAI, Anthropic, Google and other frontier language models behind one API key and one prepaid balance. It speaks the OpenAI Chat Completions and Responses formats and the Anthropic Messages format, so existing SDKs work by changing the base URL and key.

Who it is for

Developers who use more than one model vendor and want one bill
Teams moving an OpenAI or Anthropic integration without rewriting it
Apps that need streaming responses and per-call cost reporting

How it works

Create an API key in the dashboard and add credit.
Point your OpenAI or Anthropic SDK at the Omni base URL.
Send requests as usual; each response reports the amount charged in omni.charged_usd and the x-omni-charged-usd header.

Endpoints

Method	Path	What it does
`POST`	`/v1/chat/completions`	OpenAI Chat Completions, streaming supported
`POST`	`/v1/responses`	OpenAI Responses API
`POST`	`/v1/messages`	Anthropic Messages API
`GET`	`/v1/models`	Every model with its current price (no key needed)

Models

Pass one of these as model (or voice_id for voices). The full, current list with prices is at /v1/models.

gpt-6.1-sol
gpt-6-luna
claude-opus-5.5
claude-sonnet-5
gemini-3.8-flash
deepseek-v4.1-flash

Example

from openai import OpenAI

client = OpenAI(api_key="omni_live_…", base_url="https://omni-api.rcinc.app/v1")
reply = client.chat.completions.create(
    model="claude-sonnet-5",
    messages=[{"role": "user", "content": "Explain idempotency keys in two sentences."}],
)
print(reply.choices[0].message.content)
print(reply.omni["charged_usd"])  # what this call took from your credit

Pricing

Charged per token at about 25% below each vendor's public list price; for example GPT-6.1 Sol costs $1.495 per million input tokens and $7.475 per million output tokens on Omni against OpenAI's $2.00 and $10.00. Exact prices for every model are on the pricing page and at /v1/models. See the pricing page for the full comparison.

Limits

Models are served through inference marketplaces, so a vendor's newest beta features can arrive a little later than on the vendor's own API.
Vendor-specific features outside the standard request formats may need the raw passthrough endpoint.
Each key is limited to 600 requests a minute by default.