Skip to Content
DocsQuickstart

Quickstart

Get your first log in the BEval dashboard in five minutes.

1. Install

pip install bolder-ai

Requires Python 3.9+.

2. Get an API key

Open your BEval dashboard → Settings → API Keys → Create key. Copy the key (starts with bv_...).

An API key is scoped to a tenant and optionally a project. You can create multiple keys — one per service is a good default.

3. Set environment variables

export BEVAL_API_KEY=bv_...
export BEVAL_API_URL=https://ai-gateway.bolder.services   # default, optional
export BEVAL_PROJECT_ID=...                                # optional

4. Initialize and log

import beval
 
beval.init()  # reads env vars
 
beval.log(
    kind="llm",
    model_id="gpt-4o-mini",
    input="What is the capital of France?",
    output="Paris.",
    latency_ms=312,
    tokens_in=7,
    tokens_out=2,
)
 
beval.flush()  # optional — waits for the background queue to drain

Run it. Open the dashboard. The log shows up within a second.

5. Auto-wrap your existing client

If you already use OpenAI or Anthropic, skip manual logging entirely:

import beval
from openai import OpenAI
 
beval.init()
client = beval.wrap(OpenAI())
 
client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Hello in one word."}],
)

Every chat.completions.create call is now logged. Input messages, output, model, token counts, latency, errors — all captured automatically. Image parts are detected and logged as kind="vlm".

6. Trace agent functions

@beval.trace
def run_agent(query: str) -> str:
    # ...
    return answer
 
@beval.trace(name="tool:search", kind="agent")
async def search(q): ...

Captures arguments, return value, latency, and exceptions.

You’re done

Everything else in these docs is optional. Common next steps:

Last updated on
IntroductionInstallation