API Overview (Python)

Core entrypoints

create_llm(...)

Create a provider instance:

from abstractcore import create_llm

llm = create_llm("openai", model="gpt-4o-mini")  # requires: pip install "abstractcore[openai]"
resp = llm.generate("Hello!")
print(resp.content)

BasicSession

Keep conversation state:

from abstractcore import BasicSession, create_llm

session = BasicSession(create_llm("anthropic", model="claude-haiku-4-5"))
print(session.generate("Give me 3 name ideas.").content)
print(session.generate("Pick the best one.").content)

Responses (GenerateResponse)

Most calls return a GenerateResponse object (or an iterator for streaming). Common fields:

• content: cleaned assistant text
• tool_calls: structured tool calls (pass-through by default)
• usage: token usage (provider-dependent)
• metadata: provider/model-specific fields (for example extracted reasoning text when configured)

Tool calling

Tools are passed explicitly to generate() / agenerate():

from abstractcore import create_llm, tool

@tool
def get_weather(city: str) -> str:
    return f"{city}: 22°C and sunny"

llm = create_llm("openai", model="gpt-4o-mini")
resp = llm.generate("Use the tool.", tools=[get_weather])
print(resp.tool_calls)

See Tool Calling and Tool Syntax Rewriting.

Built-in tools (optional)

If you want a ready-made toolset (web + filesystem helpers), install:

pip install "abstractcore[tools]"

Then import from abstractcore.tools.common_tools (for example web_search, skim_websearch, skim_url, fetch_url). See Tool Calling for usage patterns and when to use skim_* vs fetch_*.

Structured output

Pass a Pydantic model via response_model=... to receive a typed result:

from pydantic import BaseModel
from abstractcore import create_llm

class Answer(BaseModel):
    title: str
    bullets: list[str]

llm = create_llm("openai", model="gpt-4o-mini")
result = llm.generate("Summarize HTTP/3 in 3 bullets.", response_model=Answer)
print(result.bullets)

See Structured Output.

Media input

Media handling is opt-in:

pip install "abstractcore[media]"

Then pass media=[...] to generate() / agenerate(). See Media Handling.

Audio & voice (optional)

Audio inputs are handled via an explicit policy to avoid silent semantic changes. For speech audio, use audio_policy="speech_to_text" (requires a capability plugin such as abstractvoice).

pip install abstractvoice

See Audio & Voice (STT/TTS) for usage, configuration defaults, and server endpoints.

HTTP API (optional)

If you want an OpenAI-compatible /v1 gateway, install and run the server:

pip install "abstractcore[server]"
python -m abstractcore.server.app

See HTTP Server Guide.