Source draftConnect quickstarts
Catalog checked Jun 25, 2026

Codex-style Chutes recipes

OpenAI environment mapping, provider config, authenticated smoke tests, live model picking, and sidecar fallback patterns.

Map Chutes into OpenAI env vars

Keep CHUTES_API_KEY as the durable secret name and map only inside tools that require OpenAI names.

Source
Map Chutes into OpenAI env vars
bash
export CHUTES_API_KEY=cpk_...

export OPENAI_BASE_URL="https://llm.chutes.ai/v1"
export OPENAI_API_KEY="$CHUTES_API_KEY"

Configure a provider block

Use a live model ID here. Saved default aliases require Model Routing setup first.

Source
Configure a provider block
json
{
  "provider": "openai-compatible",
  "base_url": "https://llm.chutes.ai/v1",
  "api_key_env": "CHUTES_API_KEY",
  "model": "Qwen/Qwen3-32B-TEE"
}

Run an authenticated smoke test

GET /v1/models is public, so validate the key with a completion or authenticated account endpoint.

Source
Run an authenticated smoke test
python
import os
from openai import OpenAI

client = OpenAI(
    base_url="https://llm.chutes.ai/v1",
    api_key=os.environ["CHUTES_API_KEY"],
)

response = client.chat.completions.create(
    model="Qwen/Qwen3-32B-TEE",
    messages=[{"role": "user", "content": "Confirm Chutes is configured."}],
)
print(response.choices[0].message.content)

Let a script pick the model

The picker computes presets from live data and emits inline pools that need no dashboard setup.

Source
Let a script pick the model
bash
python3 scripts/pick_model.py --task agentic --routing latency
python3 scripts/pick_model.py --need tools,structured_outputs --modality image
python3 scripts/pick_model.py --task cheap --max-input-price 0.2 --json

Which lane?

The routing pool below is computed from the live catalog for agentic work. Use it inline for zero setup, or configure a saved pool before using default:* aliases.

Concrete model
Qwen/Qwen3-32B-TEE
Use when
You need a specific context window, feature set, modality, or price.
Inline pool
Qwen/Qwen3-32B-TEE,google/gemma-4-31B-turbo-TEE,MiniMaxAI/MiniMax-M2.5-TEE:latency
Use when
You want failover or latency selection without saved alias setup.