Migration guide

What changes when you move to Inference from OpenAI, Anthropic or Google Gemini.

Migration guide

Migrating from OpenAI

Two things change: the base URL and the API key. Nothing else in your code moves.

Python

# Before
from openai import OpenAI
client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

# After
from openai import OpenAI
client = OpenAI(
    base_url="https://api.alphacurve.io/v1",
    api_key=os.environ["INFERENCE_API_KEY"],
)

Node / TypeScript

// Before
const client = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });

// After
const client = new OpenAI({
  baseURL: "https://api.alphacurve.io/v1",
  apiKey: process.env.INFERENCE_API_KEY,
});

The one catch: model IDs need a prefix

Before	After
`gpt-4o`	`openai/gpt-4o`
`gpt-4o-mini`	`openai/gpt-4o-mini`

Omitting the prefix returns 404 model_not_found.

Migrating from Anthropic

Anthropic's native API uses a different message shape. Pointing the openai SDK at Inference lets you call Claude with OpenAI-style code.

Shape mapping

Anthropic native	OpenAI-compatible (Inference)
`anthropic.Anthropic(api_key=...)`	`OpenAI(base_url="https://api.alphacurve.io/v1", api_key=...)`
`client.messages.create(...)`	`client.chat.completions.create(...)`
`system="..."` (separate parameter)	`{"role": "system", "content": "..."}` first in `messages`
`max_tokens` required	`max_tokens` optional
`model="claude-sonnet-4-5"`	`model="anthropic/claude-sonnet-4-5"`
`resp.content[0].text`	`resp.choices[0].message.content`
`resp.usage.input_tokens` / `output_tokens`	`resp.usage.prompt_tokens` / `completion_tokens`
`tools[].input_schema`	`tools[].function.parameters`
`stop_reason: "end_turn"`	`finish_reason: "stop"`

Before (Anthropic SDK)

import os
import anthropic

client = anthropic.Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])

resp = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=1024,
    system="You are concise.",
    messages=[{"role": "user", "content": "Explain routing in one sentence."}],
)
print(resp.content[0].text)

After (openai SDK pointed at Inference)

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.alphacurve.io/v1",
    api_key=os.environ["INFERENCE_API_KEY"],
)

resp = client.chat.completions.create(
    model="anthropic/claude-sonnet-4-5",
    max_tokens=1024,
    messages=[
        {"role": "system", "content": "You are concise."},
        {"role": "user", "content": "Explain routing in one sentence."},
    ],
)
print(resp.choices[0].message.content)

Migrating from Google Gemini

Gemini native	OpenAI-compatible (Inference)
`genai.GenerativeModel("gemini-2.5-pro")`	`model="google/gemini-2.5-pro"`
`contents=[{"role": "user", "parts": [...]}]`	`messages=[{"role": "user", "content": ...}]`
`role: "model"`	`role: "assistant"`
`system_instruction=`	`{"role": "system", ...}`
`generation_config.max_output_tokens`	`max_tokens`
`resp.text`	`resp.choices[0].message.content`
`inline_data` / `file_data`	`{"type": "image_url", "image_url": {"url": "data:..."}}`

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.alphacurve.io/v1",
    api_key=os.environ["INFERENCE_API_KEY"],
)

resp = client.chat.completions.create(
    model="google/gemini-2.5-pro",
    messages=[
        {"role": "system", "content": "You are concise."},
        {"role": "user", "content": "Explain routing in one sentence."},
    ],
)
print(resp.choices[0].message.content)

Migration checklist

Create an API key in the Dashboard and set a monthly spend limit.
Add credits.
Change the base URL to https://api.alphacurve.io/v1, ideally through an environment variable.
Swap in the sk-inf- key.
Add the provider/ prefix to every model ID.
Call GET /v1/models to confirm each ID exists and is available to the key.
Update error handling to read error.code (see Errors).
Canary a small slice of traffic (say 5%) before cutting over fully.
Check the Usage page in the Dashboard to confirm costs match expectations.

Supporting both during the cutover

One environment-driven configuration can serve both sides while you migrate:

import os
from openai import OpenAI

client = OpenAI(
    base_url=os.getenv("LLM_BASE_URL", "https://api.alphacurve.io/v1"),
    api_key=os.environ["LLM_API_KEY"],
)

MODEL = os.getenv("LLM_MODEL", "openai/gpt-4o-mini")

Rolling back

There is no lock-in: everything is standard OpenAI-format code. To go back to a provider directly, revert the base URL, the key, and the model prefix.