Build on Beeper.

1. Buy credits and copy your key. 2. Set BEEPER_KEY. 3. Point your SDK at the base URL.

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.beeper.ai/v1",
    api_key=os.environ["BEEPER_KEY"],
)

res = client.chat.completions.create(
    model="beeper/dolphin-mistral-24b",
    messages=[{"role": "user", "content": "Ask anything."}],
)
print(res.choices[0].message.content)

Send your key in the Authorization header. Keys are shown once — we only store a hash. Rotate them any time from Account.

Authorization: Bearer sk-beeper-XXXXXXXXXXXXXXXX

POST /v1/chat/completions — same parameters as OpenAI: model, messages, temperature, max_tokens, tools, response_format.

{
  "id": "chatcmpl-9x2…",
  "model": "beeper/hermes-4-405b",
  "upstream_model": "NousResearch/Hermes-4-405B",
  "choices": [{ "message": { "role": "assistant", "content": "…" } }],
  "usage": { "prompt_tokens": 18, "completion_tokens": 212, "cost_usd": 0.00032 }
}

Every response reports the exact upstream_model and the discounted cost_usd.

Set "stream": true to receive server-sent events. Gateway overhead is under 40 ms p50.

curl https://api.beeper.ai/v1/chat/completions -N \
  -H "Authorization: Bearer $BEEPER_KEY" -H "Content-Type: application/json" \
  -d '{"model":"beeper/dolphin-mistral-24b","stream":true,
       "messages":[{"role":"user","content":"Tell me a dark story."}]}'

POST /v1/images/generations with an unfiltered model such as beeper/flux-dev-uncensored.

{ "model": "beeper/flux-dev-uncensored", "prompt": "noir city at 3am, rain, neon", "size": "1024x1024" }

GET /v1/models returns every model with live list price and your current discounted price. Filter with ?uncensored=true.

Use Beeper as the model provider in any agent that supports OpenAI-compatible endpoints — Claude Code proxies, Cursor, Cline, OpenWebUI, SillyTavern, LiteLLM.

# LiteLLM / OpenWebUI / Cline
OPENAI_API_BASE=https://api.beeper.ai/v1
OPENAI_API_KEY=$BEEPER_KEY
MODEL=beeper/hermes-4-405b

Default 600 requests/min and 2M tokens/min per key. Limits rise automatically with lifetime spend. Headers: x-ratelimit-remaining, retry-after.