> ## Documentation Index
> Fetch the complete documentation index at: https://developers.reflection.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Deployment overview

> Run Beam yourself on your own compute, or use it through a partner

export const TokenSpeedIcon = () => <>
    <img src="/images/logos/tokenspeed.svg" alt="" className="size-6 m-0! shrink-0 dark:hidden" />
    <img src="/images/logos/tokenspeed-dark.svg" alt="" className="size-6 m-0! shrink-0 hidden dark:block" />
  </>;

Run Beam yourself: serve its open weights on your own compute, or use Beam through a cloud provider, hardware partner, or inference provider.

<CardGroup cols={3}>
  <Card title="SGLang recipe for Beam" icon="https://mintcdn.com/reflection-ai/rhhiQ-LsHnUCziYw/images/logos/sglang.svg?fit=max&auto=format&n=rhhiQ-LsHnUCziYw&q=85&s=94fda2c35da6755f6ce5a8e49ef8c235" href="https://docs.sglang.io/cookbook/autoregressive/Reflection/Beam-1_0" arrow width="640" height="640" data-path="images/logos/sglang.svg">
    The full reference for serving Beam with SGLang.
  </Card>

  <Card title="vLLM recipe for Beam" icon="https://mintcdn.com/reflection-ai/rhhiQ-LsHnUCziYw/images/logos/vllm.svg?fit=max&auto=format&n=rhhiQ-LsHnUCziYw&q=85&s=38fe2442a04f014b5454ac59dad563be" href="https://recipes.vllm.ai/reflection/Beam-1.0-501B-A23B" arrow width="24" height="24" data-path="images/logos/vllm.svg">
    The full reference for serving Beam with vLLM.
  </Card>

  <Card title="TokenSpeed recipe for Beam" icon={<TokenSpeedIcon />} href="https://lightseek.org/tokenspeed/recipes/models#beam" arrow>
    The full reference for serving Beam with LightSeek's TokenSpeed.
  </Card>
</CardGroup>

## Serve Beam on your own compute

<Steps>
  <Step title="Get the weights">
    Check the requirements and download the weights from Hugging Face. See [Download the weights](/serving-weights).
  </Step>

  <Step title="Start a server">
    Serve Beam with SGLang, vLLM, or TokenSpeed. Each guide walks through one setup on a single server. For other GPUs and precisions, use the engine recipes above.
  </Step>

  <Step title="Verify the deployment">
    Run [Beam Verifier](/beam-verifier) to check your deployment's correctness and compare its performance to reference numbers.
  </Step>
</Steps>

<CardGroup cols={3}>
  <Card title="Serve with SGLang" icon="server" href="/serving-sglang" />

  <Card title="Serve with vLLM" icon="server" href="/serving-vllm" />

  <Card title="Serve with TokenSpeed" icon="server" href="/serving-tokenspeed" />
</CardGroup>

Also for self-hosting:

* [Use HF Transformers](/cookbook/transformers): generate text, control reasoning, and call tools with Beam in Hugging Face Transformers.
* [Beam Harmony format](/harmony-format): how Beam's chat template formats conversations and how its response template parses output.
* [Speculative decoding](/speculative-decoding): speed up Beam's generation with the DFlash2 drafter.
* Quantization details: how the [FP8](/quantization-fp8) and [NVFP4](/quantization-nvfp4) weights were quantized.

## Serve Beam in production

* [Serving in production with Kubernetes](/serving-kubernetes): deploy Beam on your Kubernetes cluster with the `reflection-serving` Helm chart.
* [Scaling out with Dynamo](/serving-dynamo): scale Beam across GPUs and nodes with SGLang or vLLM and NVIDIA Dynamo.

## Use Beam through a partner

<CardGroup cols={3}>
  <Card title="Cloud providers" icon="cloud" href="/third-party-providers#cloud-providers">
    Use Beam through your cloud provider.
  </Card>

  <Card title="Hardware partners" icon="microchip" href="/third-party-providers#hardware-partners">
    Run Beam on partner hardware and systems.
  </Card>

  <Card title="Inference providers" icon="plug" href="/third-party-providers#inference-providers">
    Call Beam through third-party inference providers.
  </Card>
</CardGroup>