lm-deluge 0.0.9

Python utility for using LLM API models.

lm-deluge

lm-deluge is a lightweight helper library for maxing out your rate limits with LLM providers. It provides the following:

• Unified client – Send prompts to all relevant models with a single client.
• Massive concurrency with throttling – Set max_tokens_per_minute and max_requests_per_minute and let it fly. The client will process as many requests as possible while respecting rate limits and retrying failures.
• Spray across models/providers – Configure a client with multiple models from any provider(s), and sampling weights. The client samples a model for each request.
• Caching – Save completions in a local or distributed cache to avoid repeated LLM calls to process the same input.
• Convenient message constructor – No more looking up how to build an Anthropic messages list with images. Our Conversation and Message classes work great with our client or with the openai and anthropic packages.
• Sync and async APIs – Use the client from sync or async code.

STREAMING IS NOT IN SCOPE. There are plenty of packages that let you stream chat completions across providers. The sole purpose of this package is to do very fast batch inference using APIs. Sorry!

Installation

pip install lm-deluge

The package relies on environment variables for API keys. Typical variables include OPENAI_API_KEY, ANTHROPIC_API_KEY, COHERE_API_KEY, META_API_KEY, and GOOGLE_API_KEY. LLMClient will automatically load the .env file when imported; we recommend using that to set the environment variables.

Quickstart

The easiest way to get started is with the .basic constructor. This uses sensible default arguments for rate limits and sampling parameters so that you don't have to provide a ton of arguments.

from lm_deluge import LLMClient

client = LLMClient.basic("gpt-4o-mini")
resps = client.process_prompts_sync(["Hello, world!"])
print(resp[0].completion)

Spraying Across Models

To distribute your requests across models, just provide a list of more than one model to the constructor. The rate limits for the client apply to the client as a whole, not per-model, so you may want to increase them:

from lm_deluge import LLMClient

client = LLMClient.basic(
    ["gpt-4o-mini", "claude-haiku-anthropic"],
    max_requests_per_minute=10_000
)
resps = client.process_prompts_sync(
    ["Hello, ChatGPT!", "Hello, Claude!"]
)
print(resp[0].completion)

Configuration

API calls can be customized in a few ways.

1. Sampling Parameters. This determines things like structured outputs, maximum completion tokens, nucleus sampling, etc. Provide a custom SamplingParams to the LLMClient to set temperature, top_p, json_mode, max_new_tokens, and/or reasoning_effort. You can pass 1 SamplingParams to use for all models, or a list of SamplingParams that's the same length as the list of models. You can also pass many of these arguments directly to LLMClient.basic so you don't have to construct an entire SamplingParams object.
2. Arguments to LLMClient. This is where you set request timeout, rate limits, model name(s), model weight(s) for distributing requests across models, retries, and caching.
3. Arguments to process_prompts. Per-call, you can set verbosity, whether to display progress, and whether to return just completions (rather than the full APIResponse object).

Putting it all together:

from lm_deluge import LLMClient, SamplingParams

client = LLMClient(
    "gpt-4",
    max_requests_per_minute=100,
    max_tokens_per_minute=100_000,
    max_concurrent_requests=500,
    sampling_params=SamplingParams(temperature=0.5, max_new_tokens=30)
)

await client.process_prompts_async(
    ["What is the capital of Mars?"],
    show_progress=False,
    return_completions_only=True
)

Multi-Turn Conversations

Constructing conversations to pass to models is notoriously annoying. Each provider has a slightly different way of defining a list of messages, and with the introduction of images/multi-part messages it's only gotten worse. We provide convenience constructors so you don't have to remember all that stuff.

from lm_deluge import Message, Conversation

prompt = Conversation.system("You are a helpful assistant.").add(
    Message.user("What's in this image?").add_image("tests/image.jpg")
)

client = LLMClient.basic("gpt-4.1-mini")
resps = client.process_prompts_sync([prompt])

This just works. Images can be local images on disk, URLs, bytes, base64 data URLs... go wild. You can use Conversation.to_openai or Conversation.to_anthropic to format your messages for the OpenAI or Anthropic clients directly.

Caching

lm_deluge.cache includes LevelDB, SQLite and custom dictionary based caches. Pass an instance via LLMClient(..., cache=my_cache) and previously seen prompts will not be re‑sent across different process_prompts_[...] calls.

IMPORTANT: Caching does not currently work for prompts in the SAME batch. That is, if you call process_prompts_sync with the same prompt 100 times, there will be 0 cache hits. If you call process_prompts_sync a second time with those same 100 prompts, all 100 will be cache hits. The cache is intended to be persistent and help you save costs across many invocations, but it can't help with a single batch-inference session (yet!).

Asynchronous Client

Use this in asynchronous code, or in a Jupyter notebook. If you try to use the sync client in a Jupyter notebook, you'll have to use nest-asyncio, because internally the sync client uses async code. Don't do it! Just use the async client!

import asyncio

async def main():
    responses = await client.process_prompts_async(
        ["an async call"],
        return_completions_only=True,
    )
    print(responses[0])

asyncio.run(main())

Available Models

We support all models in src/lm_deluge/models.py. An older version of this client supported Bedrock and Vertex. We plan to re-implement Bedrock support (our previous support was spotty and we need to figure out cross-region inference in order to support the newest Claude models). Vertex support is not currently planned, since Google allows you to connect your Vertex account to AI Studio, and Vertex authentication is a huge pain (requires service account credentials, etc.)

Feature Support

We support structured outputs via json_mode parameter provided to SamplingParams. Structured outputs with a schema are planned. Reasoning models are supported via the reasoning_effort parameter, which is translated to a thinking budget for Claude/Gemini. Image models are supported. We don't support tool use yet, but support is planned (keep an eye out for a unified tool definition spec that works for all models!). We support logprobs for OpenAI models that return them via the logprobs argument to the LLMClient.

Built‑in tools

The lm_deluge.llm_tools package exposes a few helper functions:

• extract – structure text or images into a Pydantic model based on a schema.
• translate – translate a list of strings to English.
• score_llm – simple yes/no style scoring with optional log probability output.

Experimental embeddings (embed.embed_parallel_async) and document reranking (rerank.rerank_parallel_async) clients are also provided.