Installation¶

Install with your provider¶

Smelt uses LangChain under the hood, so you need the provider-specific LangChain package. Install smelt with the extra for your provider:

OpenAIAnthropicGoogle Gemini

pip install smelt-ai[openai]

Set your API key:

export OPENAI_API_KEY="sk-..."

Or pass it directly:

model = Model(provider="openai", name="gpt-4.1-mini", api_key="sk-...")

pip install smelt-ai[anthropic]

Set your API key:

export ANTHROPIC_API_KEY="sk-ant-..."

Or pass it directly:

model = Model(provider="anthropic", name="claude-sonnet-4-6", api_key="sk-ant-...")

pip install smelt-ai[google]

Set your API key:

export GOOGLE_API_KEY="..."

Or pass it directly:

model = Model(provider="google_genai", name="gemini-3-flash-preview", api_key="...")

Multiple providers¶

Install multiple extras at once:

pip install smelt-ai[openai,anthropic,google]

Using uv¶

If you use uv as your package manager:

uv add smelt-ai[openai]

Requirements¶

Python 3.10 or higher
pydantic >= 2.0
langchain >= 0.3
langchain-core >= 0.3

Verify installation¶

from smelt import Model, Job
print("smelt installed successfully")

Test that your provider is configured correctly:

from pydantic import BaseModel
from smelt import Model, Job

class TestOutput(BaseModel):
    message: str

model = Model(provider="openai", name="gpt-4.1-mini")
job = Job(prompt="Return a greeting message", output_model=TestOutput)
result = job.test(model, data=[{"input": "hello"}])
print(result.data[0])  # TestOutput(message='Hello! How can I help you today?')

Supported models¶

Any model supported by LangChain's init_chat_model works with smelt. Tested with:

Provider	`provider` value	Example models	Install extra
OpenAI	`"openai"`	`gpt-5.2`, `gpt-4.1-mini`, `gpt-4.1`, `gpt-4o`, `o4-mini`	`[openai]`
Anthropic	`"anthropic"`	`claude-sonnet-4-6`, `claude-opus-4-6`, `claude-haiku-4-5-20251001`	`[anthropic]`
Google Gemini	`"google_genai"`	`gemini-3-flash-preview`, `gemini-3-pro-preview`, `gemini-2.5-flash`	`[google]`

Using other providers

Any LangChain chat model provider works — just install the corresponding langchain-* package manually. For example, for Azure OpenAI:

pip install smelt-ai langchain-openai

model = Model(provider="azure_openai", name="my-deployment", params={
    "azure_endpoint": "https://my-resource.openai.azure.com/",
    "api_version": "2024-02-15-preview",
})

API key precedence¶

Smelt resolves API keys in this order:

Explicit api_key parameter — Model(api_key="sk-...")
Environment variable — provider-specific (e.g. OPENAI_API_KEY, ANTHROPIC_API_KEY, GOOGLE_API_KEY)
LangChain config — any provider-level configuration LangChain supports

Warning

Never hardcode API keys in source code. Use environment variables or a secrets manager (e.g. .env files with python-dotenv).