Get started

Fastest path: zero install. Or self-host in one command.

⚡ Zero install

Use the free hosted instance

No download, no signup, no key of your own to manage. Point your SDK's baseURL at the hosted mock and start testing right now:

baseURLhttps://api.llm-mock.dev/openai/v1

Anthropic lives at /anthropic, Gemini at /gemini, Gemini Enterprise at /gemini-enterprise and Azure at /azure/openai on the same host.

Prefer to self-host?

Prebuilt multi-arch image on GHCR, or run from source with Bun.

# Docker
docker run --rm -p 3000:3000 ghcr.io/axium-lab/llm-mock

# From source
git clone https://github.com/axium-lab/llm-mock.git
cd llm-mock
bun install
bun start

See the API reference for every endpoint, header and error shape — one page per provider: OpenAI, Anthropic, Gemini, Gemini Enterprise, Azure OpenAI.

Change one line, keep your code

The official openai, @anthropic-ai/sdk and @google/genai SDKs talk to llm-mock exactly as they talk to the real APIs. No mocking libraries, no request interception.

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "http://localhost:3000/openai/v1", // the only change in your app
  apiKey: "sk-mock-key-01",                   // any mock key, no real secret
});

const completion = await client.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "Hello!" }],
});

console.log(completion.choices[0].message.content); // "Echo: Hello!"

Same idea for Anthropic — its client appends only the request path, so the version segment belongs in the baseURL:

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: "sk-mock-key-01",
  baseURL: "http://localhost:3000/anthropic",
});

const message = await client.messages.create({
  model: "claude-opus-5",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello!" }],
});

console.log(message.content[0].text); // "Echo: Hello!"

And for Gemini — note that @google/genai appends the version segment itself, so its baseUrl stops at the prefix:

import { GoogleGenAI } from "@google/genai";

const ai = new GoogleGenAI({
  apiKey: "sk-mock-key-01",
  httpOptions: { baseUrl: "http://localhost:3000/gemini" },
});

const response = await ai.models.generateContent({
  model: "gemini-3.6-flash",
  contents: "Hello!",
});

console.log(response.text); // "Echo: Hello!"

Built for real integration tests

Not a stub — a faithful contract, verified in CI against each provider's official SDK.

Faithful contracts

Responses match each provider's real shapes, validated in CI with the official openai, @anthropic-ai/sdk and @google/genai SDKs as the clients.

Five providers, one server

OpenAI, Anthropic, Gemini (AI Studio), Gemini Enterprise and Azure OpenAI, each under its own prefix with its own auth scheme and error envelope.

Streaming (SSE)

stream: true works everywhere, in all three conventions the five providers use — including Anthropic's named event: frames.

Deterministic & stateless

Same request, same bytes. Snapshot-friendly, and identical on a laptop, in CI, or behind a load balancer.

Tool calls

Real tool_calls with arguments synthesized from your JSON Schema, streaming included — so agent loops run end to end.

Files & uploads

Multipart uploads and the chunked Uploads API work, yet nothing is persisted: the id itself carries the file's metadata.

Real error flows

Invalid keys, unknown models, validation errors and content filters return each provider's exact error envelope — test your error handling too.

Controllable replies

Need a specific answer? Send it in a request header. Nothing to register, nothing to clean up.

Zero setup

One Docker command, or bun install && bun start. Mock keys ship in the repo.

Provider support

Multi-provider by design: each provider mounts under its own URL prefix and implements its own API contract. Here's where each one stands today.

Provider	Prefix	Status
OpenAI	`/openai/v1`	✅ Supported
Anthropic	`/anthropic`	✅ Supported
Gemini (AI Studio)	`/gemini`	✅ Supported
Gemini Enterprise (ex-Vertex AI)	`/gemini-enterprise`	✅ Supported
Azure OpenAI	`/azure/openai`	✅ Supported