langchain-pi

Native LangChain chat models for OpenAI Codex (ChatGPT Plus/Pro subscription OAuth), Claude (Claude Code subscription OAuth), and OpenCode Zen/Go. No Pi runtime, no Node sidecar.

Requirements

• Python >= 3.9
• For Codex: an openai-codex credential in ~/.pi/agent/auth.json, or sign in with codex-login
• For Claude Code: the Claude Code CLI logged in once (~/.claude/.credentials.json)
• For paid OpenCode models: OPENCODE_API_KEY or an explicit api_key

Install

pip install langchain-pi
# or: uv add langchain-pi

Usage

create_chat routes the supported native providers:

from langchain_pi import create_chat

codex = create_chat("openai-codex", "gpt-5.3-codex-spark")
claude = create_chat("claude-code", "claude-sonnet-4-6")
free = create_chat("opencode", "deepseek-v4-flash-free")
go = create_chat("opencode-go", "glm-5", api_key="...")

Or construct Codex directly:

from langchain_pi import ChatCodex

model = ChatCodex(
    model="gpt-5.3-codex-spark",
    reasoning="minimal",
    system="You are a helpful assistant.",
)

print(model.invoke("Hello!").content)

Codex Auth

ChatCodex reads the same ~/.pi/agent/auth.json credential shape under openai-codex, refreshes the OAuth token in place, and talks directly to https://chatgpt.com/backend-api/codex/responses.

If no credential exists:

codex-login
codex-login --device

Tool Calling

from langchain_core.tools import tool
from langchain_pi import ChatCodex

@tool
def get_weather(city: str) -> str:
    """Get the current weather for a city."""
    return f"It is sunny in {city}, 24C."

model = ChatCodex(model="gpt-5.3-codex").bind_tools([get_weather])
msg = model.invoke("What's the weather in Paris?")
print(msg.tool_calls)

tool_choice is passed through to the Codex Responses API:

forced = ChatCodex(model="gpt-5.3-codex").bind_tools(
    [get_weather],
    tool_choice={"type": "function", "name": "get_weather"},
)

For agent loops, keep tool_choice="auto" unless every model turn should call the same tool.

Streaming

for chunk in model.stream("Write a haiku."):
    print(chunk.content, end="")

OpenCode

ChatOpencode uses OpenCode's OpenAI-compatible endpoints through langchain-openai.

from langchain_pi import ChatOpencode

free = ChatOpencode("deepseek-v4-flash-free")
paid = ChatOpencode("glm-5")
go = ChatOpencode("glm-5", tier="go")

Free models include deepseek-v4-flash-free, big-pickle, mimo-v2.5-free, and nemotron-3-super-free.

Claude Code (Anthropic subscription)

ChatClaudeCode talks to the Anthropic Messages API authenticated with the Claude Code OAuth session already on the machine (~/.claude/.credentials.json), billing requests against your Claude Code subscription — no API key. It reads and refreshes the token in place (with a claude CLI fallback).

from langchain_pi import ChatClaudeCode, create_chat

chat = create_chat("claude-code", "claude-sonnet-4-6")
print(chat.invoke("Hello!").content)

# Or construct directly, with options:
opus = ChatClaudeCode(model="claude-opus-4-8", reasoning="medium")

Models: claude-opus-4-8, claude-opus-4-7, claude-sonnet-4-6, claude-haiku-4-5. Reasoning uses adaptive thinking on Opus 4.8/4.7 and a token budget on Sonnet 4.6 (Haiku has no reasoning). The 1M-context beta is opt-in via long_context=True; the subscription rejects long-context requests without extra credits otherwise. Tool calling and streaming work as with ChatCodex.

Using a subscription OAuth session from a third-party app may violate Anthropic's terms and risk your account. See the pi-claude-code-auth README before relying on this.

License

MIT