OpenHands

What you need

• Base URLhttps://api.opencheap.si/v1
• API keyA kl_live_… key from the console, kept in OPENCHEAP_API_KEY.
• ModelAn id from Models, for example claude-sonnet-5.

In the app

1. 1
   Open Settings, then the LLM tab, and turn on Advanced.
2. 2
   Set Custom Model.

   Type openai/claude-sonnet-5. The openai/ is LiteLLM's prefix for an OpenAI-compatible endpoint; the id after it is the OpenCheap model.
3. 3
   Set Base URL.

   https://api.opencheap.si/v1
4. 4
   Paste your key into API Key.

   Or create a Provider Connection with the key and base URL once, and pick it in the profile's Provider Connection field.
5. 5
   Save, start a new conversation and send a short message.

With the Agent SDK

import os
from pydantic import SecretStr
from openhands.sdk import LLM, Message, TextContent

llm = LLM(
    model="openai/claude-sonnet-5",
    base_url="https://api.opencheap.si/v1",
    api_key=SecretStr(os.environ["OPENCHEAP_API_KEY"]),
    usage_id="opencheap",
)
resp = llm.completion(messages=[Message(role="user", content=[TextContent(text="Say hello in five words.")])])
print(resp.message.content[0].text)

Install it with pip install openhands-sdk. The same LLM goes into an Agent as usual. To configure it from the environment instead, set LLM_MODEL, LLM_BASE_URL and LLM_API_KEY and call LLM.load_from_env().

The SDK may print "Cost calculation failed: This model isn't mapped yet". It is LiteLLM not knowing the model's price; the call is not affected, and your usage and spend are in the console.

What was checked

Run on October 28, 2026 with openhands-sdk 1.58.2 on claude-sonnet-5: llm.completion as above, and again configured only through LLM_MODEL, LLM_BASE_URL and LLM_API_KEY. The app steps and the Custom Model, Base URL and API Key labels are from the current OpenHands settings source and documentation; the app itself was not run.