prune-sdk 0.1.0

Reduce LLM API costs by 40-70% with one line of code — drop-in Anthropic/OpenAI client via the Prune proxy

Prune SDK

Reduce your LLM API costs by 40–70% (blended over repeat & similar traffic) with a one-line import change — no prompt edits, same responses on cache miss.

# Before
from anthropic import Anthropic

# After
from prune import Anthropic

Supported providers

Provider	Status
Anthropic (Claude Opus, Sonnet, Haiku)	Live via proxy
OpenAI (GPT-4o, GPT-4o-mini, o-series)	Live via proxy
Google Gemini	Planned

Installation

pip install prune-sdk

For local backend development:

export PRUNE_BASE_URL="http://127.0.0.1:8000"

Install from source (contributors):

pip install -e "./prune-sdk[dev]"

Quick start

Anthropic (Claude)

from prune import Anthropic

client = Anthropic(
    api_key="sk-ant-your-key",
    prune_api_key="prune_your_key",
)

message = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello, Claude!"}],
)

print(message.content[0].text)
print(client.last_prune_metadata)  # cache hit, tokens saved, etc.

OpenAI (GPT)

from prune import OpenAI

client = OpenAI(
    api_key="sk-your-openai-key",
    prune_api_key="prune_your_key",
)

completion = client.chat.completions.create(
    model="gpt-4o-mini",
    max_tokens=256,
    messages=[{"role": "user", "content": "Hello!"}],
)

print(completion.choices[0].message.content)

Async

from prune import AsyncAnthropic

client = AsyncAnthropic(
    api_key="sk-ant-...",
    prune_api_key="prune_...",
)

message = await client.messages.create(
    model="claude-3-5-haiku-20241022",
    max_tokens=100,
    messages=[{"role": "user", "content": "Hi"}],
)

Configuration

Environment variables:

export PRUNE_API_KEY="prune_your_key"
export PRUNE_BASE_URL="https://api.prune.so"   # or http://127.0.0.1:8000 for local backend
export PRUNE_FALLBACK="true"                     # fallback to direct API if proxy fails

Programmatic:

import prune

prune.configure(api_key="prune_your_key", base_url="https://api.prune.so")

client = prune.Anthropic(api_key="sk-ant-...")

Behavior

Feature	Details
Proxy routing	Anthropic → /v1/proxy/anthropic/messages · OpenAI → /v1/proxy/openai/chat/completions
Quality	Cache miss = same payload to the provider as without Prune. Cache hit = identical prior response.
Savings	Exact + semantic cache; Claude system prompt caching. See docs/SAVINGS_MODEL.md.
Response type	Real anthropic.types.Message / ChatCompletion objects
Streaming	Bypasses Prune; uses official SDK directly
Fallback	On proxy outage (5xx / network), calls Anthropic/OpenAI directly
Disable Prune	Anthropic(..., enable_prune=False)
Prompt Pass	HTTP header X-Prune-Optimize: light or compact (optional; default off)

Direct HTTP (no SDK)

If you only need to test the proxy, skip the SDK and POST to the backend:

curl -X POST http://127.0.0.1:8000/v1/proxy/anthropic/messages ^
  -H "X-Prune-Key: prune_your_key" ^
  -H "Content-Type: application/json" ^
  -d "{\"model\":\"claude-sonnet-4-20250514\",\"max_tokens\":64,\"messages\":[{\"role\":\"user\",\"content\":\"Hello\"}],\"user_api_key\":\"sk-ant-...\"}"

import httpx

resp = httpx.post(
    "http://127.0.0.1:8000/v1/proxy/anthropic/messages",
    headers={"X-Prune-Key": "prune_your_key"},
    json={
        "model": "claude-sonnet-4-20250514",
        "max_tokens": 64,
        "messages": [{"role": "user", "content": "Hello"}],
        "user_api_key": "sk-ant-...",
    },
    timeout=60,
)
print(resp.json())

Development

cd prune-sdk
pip install -e ".[dev]"
pytest tests/ -q
pytest tests/ -m integration  # needs ANTHROPIC_API_KEY + PRUNE_API_KEY

License

MIT