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

# Beam 1.0

> Deploy Reflection Beam 1.0 with SGLang on B300, GB300, and H200: NVFP4, FP8, and BF16 recipes for low latency and high throughput.

## Deployment

> **Install SGLang**
>
> ```bash Command theme={null}
> docker pull lmsysorg/sglang:beam
> ```
>
> For host setup, see [Install SGLang with Docker](https://docs.sglang.io/docs/get-started/install.md#method-3-using-docker). The command panel below has a **Python / Docker** toggle, with Python selected by default. The Python output is the bare `sglang serve` command: run it inside a container started from this image. The Docker output wraps the same command in `docker run`.

Select your hardware, strategy, and quantization to generate the launch command.

* **Low-Latency** uses TP with the [DFlash2 drafter](https://huggingface.co/reflection/Beam-1.0-501B-A23B-DFlash2) at block size 8.
* **High-Throughput** disables DFlash2. H200 BF16 uses TP8 with an automatic static memory fraction; the other recipes use TP + DPA. Blackwell NVFP4 also enables EP and FlashInfer MegaMoE.

Each GSM8K score covers all 1,319 test problems; the benchmark notes identify the evaluation harness.

NVFP4 requires B300 or GB300. **GB300 BF16 uses two four-GPU nodes**: set the head-node address and node rank in **Env**, then run the command on both nodes with rank 0 on the head and rank 1 on the peer.

## TPU Support

Beam 1.0 has day-0 TPU support through SGLang-TorchTPU, a native PyTorch backend for LLM, VLM, and diffusion model inference on TPUs. It combines TPU acceleration with SGLang features such as RadixAttention and HiCache to enable cost-efficient, production-ready serving for agentic workloads.

TPU recipes for Beam 1.0 and access to SGLang-TorchTPU will be available soon. Sign up through this [Google Form](https://docs.google.com/forms/d/e/1FAIpQLSe2Y8hzyU3KGB2StA-0huulrVsq79A0tSxcBQwjj8vlhLhYuQ/viewform?usp=publish-editor) to receive the latest updates.

## Usage

All recipes serve the model as `beam` on port 30000. For GB300 BF16, connect to the head node.

> **Chat completion (Python)**
>
> ```python Example theme={null}
> from openai import OpenAI
>
> client = OpenAI(base_url="http://localhost:30000/v1", api_key="EMPTY")
> response = client.chat.completions.create(
>     model="beam",
>     messages=[{"role": "user", "content": "Explain tensor parallelism briefly."}],
> )
> message = response.choices[0].message
> print("Reasoning:", message.reasoning_content)
> print("Answer:", message.content)
> ```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.