Docs
Integrations
Point editors, SDKs, and OpenAI/Anthropic clients at your local Bedrova server.
Anything that speaks the OpenAI or Anthropic API can use Bedrova. The rule is the same every
time: set the base URL to Bedrova’s root (http://127.0.0.1:1337), use any placeholder
key on localhost, and pick a model id from /v1/models. Ready-made config
for the common clients is below.
OpenAI SDKs
Python:
from openai import OpenAI
client = OpenAI(base_url="http://127.0.0.1:1337/v1", api_key="local")
print(client.chat.completions.create(
model="qwen35-7b-4bit",
messages=[{"role": "user", "content": "Say hi"}],
).choices[0].message.content)
Node:
import OpenAI from "openai";
const client = new OpenAI({ baseURL: "http://127.0.0.1:1337/v1", apiKey: "local" });
const r = await client.chat.completions.create({
model: "qwen35-7b-4bit",
messages: [{ role: "user", content: "Say hi" }],
});
console.log(r.choices[0].message.content);
Anthropic SDK
from anthropic import Anthropic
client = Anthropic(base_url="http://127.0.0.1:1337/anthropic", api_key="local")
msg = client.messages.create(
model="qwen35-7b-4bit",
max_tokens=256,
messages=[{"role": "user", "content": "Say hi"}],
)
print(msg.content)
Editors and coding tools
Tools that let you set a custom OpenAI-compatible endpoint (Continue.dev, editor AI plugins, Claude-compatible clients) work by entering the base URL and a model id:
- Base URL:
http://127.0.0.1:1337/v1(OpenAI dialect) orhttp://127.0.0.1:1337/anthropic(Anthropic dialect). - API key: any non-empty placeholder on localhost.
- Model: an installed id from
/v1/models.
Pin the model you code against so it stays resident and responds without a load delay — see Memory & loading.
Using Bedrova as a local backend
If another app expects a local OpenAI/Anthropic provider, give it the root URL
(http://127.0.0.1:1337), not a trailing /v1. Bedrova exposes rich metadata at
/api/v1/models, so well-behaved clients can discover what you have
loaded.
Exposing to other devices
To reach Bedrova from another machine, enable LAN/Tailscale exposure in
Settings → Server/API and set an access key. Until you do, the
server only answers on 127.0.0.1. The trade-offs are covered in
Network calls.