Authentication

One DirectInference key authenticates every surface. Only the header convention differs, matching what each SDK already sends.

API keys

Authenticate with a key you issue on the API Keys page. The same llm_live_… key works across the OpenAI, Anthropic, and Gemini surfaces — there are no per-surface credentials.

Surface	Credential header(s)
OpenAI-compatible	`Authorization: Bearer <key>`
Anthropic Messages	`x-api-key: <key>` `anthropic-version: 2023-06-01`
Gemini	`x-goog-api-key: <key>` `?key=<key>` `Authorization: Bearer <key>`

Base URLs

Every surface lives under the /di prefix — the DirectInference endpoint, where each request is classified, your effort hint is honored, and quality repair runs. Set your client’s base URL and change nothing else:

Surface	Base URL
OpenAI-compatible	`https://api.directinference.com/di/v1`
Anthropic Messages	`https://api.directinference.com/di` the SDK appends `/v1/messages`
Gemini	`https://api.directinference.com/di` the SDK appends `/v1beta/…`; Vercel `@ai-sdk/google` wants `…/di/v1beta`

OpenAI-compatible

Pass the key as api_key; the SDK sends it as a bearer token.

from openai import OpenAI

client = OpenAI(
    api_key="llm_live_...",                       # sent as: Authorization: Bearer ...
    base_url="https://api.directinference.com/di/v1",
)

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "llm_live_...",                         // sent as: Authorization: Bearer ...
  baseURL: "https://api.directinference.com/di/v1",
});

curl https://api.directinference.com/di/v1/chat/completions \
  -H "Authorization: Bearer llm_live_..." \
  -H "Content-Type: application/json" \
  -d '{ "model": "gpt-5.5-mini", "messages": [{ "role": "user", "content": "ping" }] }'

echo '{ "model": "gpt-5.5-mini", "messages": [{ "role": "user", "content": "ping" }] }' \
  | https POST api.directinference.com/di/v1/chat/completions \
      Authorization:'Bearer llm_live_...' \
      Content-Type:application/json

client := openai.NewClient(
  option.WithAPIKey("llm_live_..."),            // sent as: Authorization: Bearer ...
  option.WithBaseURL("https://api.directinference.com/di/v1"),
)

Anthropic Messages

The Anthropic SDK sends x-api-key plus an anthropic-version header. Set the SDK base URL to …/di (it appends /v1/messages for you).

curl
HTTPie

curl https://api.directinference.com/di/v1/messages \
  -H "x-api-key: llm_live_..." \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-6",
    "max_tokens": 64,
    "messages": [{ "role": "user", "content": "ping" }]
  }'

echo '{ "model": "claude-sonnet-4-6", "max_tokens": 64, "messages": [{ "role": "user", "content": "ping" }] }' \
  | https POST api.directinference.com/di/v1/messages \
      x-api-key:llm_live_... \
      anthropic-version:2023-06-01 \
      Content-Type:application/json

Gemini

Three credential forms are accepted, resolved in this order: x-goog-api-key, then the ?key= query parameter, then Authorization: Bearer. The Google GenAI SDKs use the first.

curl
HTTPie

# x-goog-api-key is what the Google GenAI SDKs send
curl "https://api.directinference.com/di/v1beta/models/gemini-2.5-flash:generateContent" \
  -H "x-goog-api-key: llm_live_..." \
  -H "Content-Type: application/json" \
  -d '{ "contents": [{ "parts": [{ "text": "ping" }] }] }'

# also accepted: ?key=llm_live_...   and   Authorization: Bearer llm_live_...

# x-goog-api-key is what the Google GenAI SDKs send
echo '{ "contents": [{ "parts": [{ "text": "ping" }] }] }' \
  | https POST 'api.directinference.com/di/v1beta/models/gemini-2.5-flash:generateContent' \
      x-goog-api-key:llm_live_... \
      Content-Type:application/json

# also accepted: ?key=llm_live_...   and   Authorization:'Bearer llm_live_...'