Model Gateway¶

ASHA v0.4.2 - adapters and client wrapping.

wrap_llm (recommended)¶

from asha.integrations import wrap_llm
import openai

client = wrap_llm(openai.OpenAI(), mode="balanced")

response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Data from john@example.com"}],
)

Import from asha.integrations, not root.

Providers¶

Provider	Extra	Env var
OpenAI	`asha[openai]`	`OPENAI_API_KEY`
Anthropic	`asha[anthropic]`	`ANTHROPIC_API_KEY`
Gemini	`asha[gemini]`	`GOOGLE_API_KEY`
Grok	-	`GROK_API_KEY`
Ollama	-	local server
HuggingFace	`asha[transformers]`	-
Mock	-	testing only

Agent¶

End-to-end preprocess + generate:

from asha import Agent

agent = Agent(model="gpt-4o-mini", privacy=True)
response = agent.run("Analyze Q1 sales")

Mock adapter (no API key):

agent = Agent(model="mock")

AdapterFactory¶

from asha.runtime.adapters.factory import AdapterFactory

adapter = AdapterFactory.create(provider="openai", model="gpt-4o-mini")
adapter = AdapterFactory.create(provider="mock")

Smart routing¶

from asha import Agent

agent = Agent(
    routing_config={
        "chat": "gpt-4o-mini",
        "analysis": "gpt-4o",
    }
)
agent.run("Summarize", task_type="chat")

See routing.md.

auto_patch¶

from asha.integrations import auto_patch, disable_auto_patch

auto_patch(mode="strict")
disable_auto_patch()

Global SDK patch - prefer wrap_llm() for production.