anymodel-py 0.1.0

OpenRouter-compatible LLM router with unified batch support. Route requests across OpenAI, Anthropic, and Google with a single API.

anymodel

OpenRouter-compatible LLM router with unified batch support for Python. Self-hosted, zero fees.

Route requests across OpenAI, Anthropic, and Google with a single API. Add any OpenAI-compatible provider. Run as an SDK or standalone HTTP server.

Install

pip install anymodel

Quick Start

Set your API keys as environment variables:

export OPENAI_API_KEY=sk-...
export ANTHROPIC_API_KEY=sk-ant-...
export GOOGLE_API_KEY=AIza...

SDK Usage

import asyncio
from anymodel import AnyModel

async def main():
    client = AnyModel()

    response = await client.chat.completions.create(
        model="anthropic/claude-sonnet-4-6",
        messages=[{"role": "user", "content": "Hello!"}],
    )
    print(response["choices"][0]["message"]["content"])

asyncio.run(main())

Streaming

stream = await client.chat.completions.create(
    model="openai/gpt-4o",
    messages=[{"role": "user", "content": "Write a haiku"}],
    stream=True,
)

async for chunk in stream:
    content = chunk["choices"][0].get("delta", {}).get("content", "")
    print(content, end="", flush=True)

Supported Providers

Set the env var and go. Models are auto-discovered from each provider's API.

Provider	Env Var	Example Model
OpenAI	OPENAI_API_KEY	openai/gpt-4o
Anthropic	ANTHROPIC_API_KEY	anthropic/claude-sonnet-4-6
Google	GOOGLE_API_KEY	google/gemini-2.5-pro
Mistral	MISTRAL_API_KEY	mistral/mistral-large-latest
Groq	GROQ_API_KEY	groq/llama-3.3-70b-versatile
DeepSeek	DEEPSEEK_API_KEY	deepseek/deepseek-chat
xAI	XAI_API_KEY	xai/grok-3
Together	TOGETHER_API_KEY	together/meta-llama/Llama-3.3-70B-Instruct-Turbo
Fireworks	FIREWORKS_API_KEY	fireworks/accounts/fireworks/models/llama-v3p3-70b-instruct
Perplexity	PERPLEXITY_API_KEY	perplexity/sonar-pro
Ollama	OLLAMA_BASE_URL	ollama/llama3.3

Fallback Routing

Try multiple models in order. If one fails, the next is attempted:

response = await client.chat.completions.create(
    model="",
    models=[
        "anthropic/claude-sonnet-4-6",
        "openai/gpt-4o",
        "google/gemini-2.5-pro",
    ],
    route="fallback",
    messages=[{"role": "user", "content": "Hello"}],
)

Tool Calling

response = await client.chat.completions.create(
    model="anthropic/claude-sonnet-4-6",
    messages=[{"role": "user", "content": "What's the weather in NYC?"}],
    tools=[{
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "Get current weather for a location",
            "parameters": {
                "type": "object",
                "properties": {"location": {"type": "string"}},
                "required": ["location"],
            },
        },
    }],
    tool_choice="auto",
)

for call in response["choices"][0]["message"].get("tool_calls", []):
    print(call["function"]["name"], call["function"]["arguments"])

Batch Processing

Process many requests with native provider batch APIs or concurrent fallback.

Submit and wait

results = await client.batches.create_and_poll({
    "model": "openai/gpt-4o-mini",
    "requests": [
        {"custom_id": "req-1", "messages": [{"role": "user", "content": "Summarize AI"}]},
        {"custom_id": "req-2", "messages": [{"role": "user", "content": "Summarize ML"}]},
    ],
})

for result in results["results"]:
    print(result["custom_id"], result["response"]["choices"][0]["message"]["content"])

Submit now, check later

# Submit and get the batch ID
batch = await client.batches.create({
    "model": "anthropic/claude-haiku-4-5",
    "requests": [
        {"custom_id": "req-1", "messages": [{"role": "user", "content": "Summarize AI"}]},
    ],
})
print(batch["id"])  # "batch-abc123"

# Check status any time
status = await client.batches.get("batch-abc123")
print(status["status"])  # "pending", "processing", "completed"

# Wait for results when ready
results = await client.batches.poll("batch-abc123")

# List all batches
all_batches = await client.batches.list()

# Cancel a batch
await client.batches.cancel("batch-abc123")

Batch configuration

client = AnyModel({
    "batch": {
        "poll_interval": 10.0,          # default poll interval in seconds
        "concurrency_fallback": 10,      # concurrent request limit for non-native providers
    },
    "io": {
        "read_concurrency": 30,          # concurrent file reads (default: 20)
        "write_concurrency": 15,         # concurrent file writes (default: 10)
    },
})

Configuration

client = AnyModel({
    "anthropic": {"api_key": "sk-ant-..."},
    "openai": {"api_key": "sk-..."},
    "aliases": {
        "default": "anthropic/claude-sonnet-4-6",
        "fast": "anthropic/claude-haiku-4-5",
        "smart": "anthropic/claude-opus-4-6",
    },
    "defaults": {
        "temperature": 0.7,
        "max_tokens": 4096,
        "retries": 2,
    },
})

# Use aliases as model names
response = await client.chat.completions.create(
    model="fast",
    messages=[{"role": "user", "content": "Quick answer"}],
)

Config File

Create anymodel.config.json in your project root:

{
  "anthropic": {
    "api_key": "${ANTHROPIC_API_KEY}"
  },
  "aliases": {
    "default": "anthropic/claude-sonnet-4-6"
  },
  "defaults": {
    "temperature": 0.7,
    "max_tokens": 4096
  }
}

${ENV_VAR} references are interpolated from environment variables.

Custom Providers

Add any OpenAI-compatible endpoint:

client = AnyModel({
    "custom": {
        "ollama": {
            "base_url": "http://localhost:11434/v1",
            "models": ["llama3.3", "mistral"],
        },
    },
})

response = await client.chat.completions.create(
    model="ollama/llama3.3",
    messages=[{"role": "user", "content": "Hello from Ollama"}],
)

Server Mode

Run as a standalone HTTP server compatible with the OpenAI SDK:

pip install anymodel[server]
anymodel serve --port 4141

Then point any OpenAI-compatible client at it:

from openai import OpenAI

client = OpenAI(base_url="http://localhost:4141/api/v1", api_key="unused")
response = client.chat.completions.create(
    model="anthropic/claude-sonnet-4-6",
    messages=[{"role": "user", "content": "Hello via server"}],
)

Also Available

• Node.js: @probeo/anymodel on npm

License

MIT