Beam

Beam

Reflection
+
+

Related Products

  • LTX
    182 Ratings
    Visit Website
  • Cloverleaf
    189 Ratings
    Visit Website
  • Portfolio Manager
    3 Ratings
    Visit Website
  • Interfacing Integrated Management System (IMS)
    66 Ratings
    Visit Website
  • GWI
    201 Ratings
    Visit Website
  • InEight
    168 Ratings
    Visit Website
  • JetBrains Junie
    12 Ratings
    Visit Website
  • Runpod
    230 Ratings
    Visit Website
  • Fraud.net
    56 Ratings
    Visit Website
  • Dialpad Support
    1,600 Ratings
    Visit Website

About

Beam is Reflection’s first open-weight model, a sparse Mixture-of-Experts model with 501 billion total parameters and 23 billion active parameters, built for coding, reasoning, and agentic workloads. Its capabilities come from large-scale pretraining and reinforcement learning, including training on 23.8 trillion diverse, curated, high-quality tokens from the web, public sources, and proprietary licensed datasets. Beam was trained with a particular focus on coding and agentic performance and is designed to deliver competitive open-weight capabilities with efficient inference compute. It supports complex software engineering, terminal, STEM, web search, tool-use, and general knowledge tasks, with reinforcement learning designed to improve multi-step reasoning, tool use, and adaptation to environment feedback. Users can control the tradeoff between performance and token usage through a reasoning effort parameter.

About

Qwen3.8-Flash-Next is an open-weight multimodal Mixture-of-Experts model and an early preview of the architecture planned for Qwen4. It systematically upgrades attention, residual connections, embeddings, and optimization to improve capability, computational efficiency, model capacity, and training stability. Its hybrid architecture combines Gated DeltaNet, which efficiently compresses historical information, with Qwen Sparse Attention, which selects important context at the micro-block level to reduce attention and indexing costs on long sequences. Gated Residual widens the residual stream into four branches and dynamically controls information flow across layers, while N-gram Embedding adds large-scale local-pattern memory with very little extra per-token computation and can be offloaded to host memory. The model uses a 125B-parameter main network plus 51B N-gram embedding parameters, while activating only 6B parameters per token.

Platforms Supported

Windows Supported
Mac Supported
Linux Supported
Cloud Supported
On-Premises Supported
iPhone Not Supported
iPad Not Supported
Android Not Supported
Chromebook Not Supported

Platforms Supported

Windows Not Supported
Mac Not Supported
Linux Not Supported
Cloud Supported
On-Premises Not Supported
iPhone Not Supported
iPad Not Supported
Android Not Supported
Chromebook Not Supported

Audience

Developers, AI researchers, and engineering teams requiring a tool to run coding, reasoning, tool-use, and agentic workloads with an efficient open-weight language model

Audience

Developers, researchers, and AI teams seeking to run or study an efficient multimodal open-weight model with long-context reasoning, coding, multilingual, and agentic capabilities

Support

Phone Support Not Supported
24/7 Live Support Not Supported
Online Supported

Support

Phone Support Not Supported
24/7 Live Support Not Supported
Online Supported

API

Offers API Not Supported

API

Offers API Supported

Screenshots and Videos

Screenshots and Videos

Pricing

No information available.
Free Version Not Supported
Free Trial Not Supported

Pricing

$2 per 1M (input)
Free Version Not Supported
Free Trial Not Supported

Reviews/Ratings

Overall 0.0 / 5
ease 0.0 / 5
features 0.0 / 5
design 0.0 / 5
support 0.0 / 5

This software hasn't been reviewed yet. Be the first to provide a review:

Review this Software

Reviews/Ratings

Overall 0.0 / 5
ease 0.0 / 5
features 0.0 / 5
design 0.0 / 5
support 0.0 / 5

This software hasn't been reviewed yet. Be the first to provide a review:

Review this Software

Training

Documentation Supported
Webinars Not Supported
Live Online Supported
In Person Not Supported

Training

Documentation Supported
Webinars Not Supported
Live Online Not Supported
In Person Not Supported

Company Information

Reflection
United States
reflection.ai/blog/introducing-beam

Company Information

Alibaba
Founded: 1999
China
qwen.ai/blog

Alternatives

GLM-5.3

GLM-5.3

Z.ai

Alternatives

Kimi K3

Kimi K3

Moonshot AI
Qwen3.8-Max

Qwen3.8-Max

Alibaba
Inkling

Inkling

Thinking Machines Lab
Qwen3.5

Qwen3.5

Alibaba
Qwen3.5

Qwen3.5

Alibaba

Categories

AI Models Supported

Categories

AI Coding Models Supported
AI Models Supported
Foundation Models Supported
Multimodal Models Supported

Integrations

Alibaba Cloud Not Supported
Alibaba Cloud Model Studio Not Supported
Cherry Studio Not Supported
Cline Not Supported
ClinePass Not Supported
Happy Shrimp 1.0 Not Supported
Hermes Agent Not Supported
Hugging Face Not Supported
Model Context Protocol (MCP) Not Supported
ModelScope Not Supported
Odysseus Not Supported
OfoxAI Not Supported
Ollama Not Supported
OpenClaw Not Supported
Python Not Supported
Qwen Not Supported
Qwen Code Not Supported
Qwen Studio Not Supported
QwenCloud Not Supported
QwenWork Not Supported

Integrations

Alibaba Cloud Supported
Alibaba Cloud Model Studio Supported
Cherry Studio Supported
Cline Supported
ClinePass Supported
Happy Shrimp 1.0 Supported
Hermes Agent Supported
Hugging Face Supported
Model Context Protocol (MCP) Supported
ModelScope Supported
Odysseus Supported
OfoxAI Supported
Ollama Supported
OpenClaw Supported
Python Supported
Qwen Supported
Qwen Code Supported
Qwen Studio Supported
QwenCloud Supported
QwenWork Supported
Claim Beam and update features and information
Claim Beam and update features and information
Claim Qwen3.8-Flash-Next and update features and information
Claim Qwen3.8-Flash-Next and update features and information