DBRX
by Databricks · www.databricks.com ↗
Open-weight mixture-of-experts LLM released by Databricks in 2024; retired from Databricks serving in 2025, though the weights remain available.
open-weights llm mixture-of-experts retired
- Category
- Open Models
- Business model
- Open (open weights/source)
- Availability
- Global
- Launched
- 2024-03
- Record updated
- 2026-07-30
- Model family
- DBRX
- License
- Databricks Open Model License
- Founded
- 2013
- Headquarters
- US
- Canonical URL
- https://globalaiproductindex.com/products/dbrx/
Overview
DBRX is an open-weight mixture-of-experts large language model released by Databricks in March 2024, with 132 billion total parameters of which 36 billion are active per token. At release it led open-model benchmarks in programming and general knowledge. Databricks retired DBRX from its own model serving in 2025, but the weights remain downloadable under the Databricks Open Model License.
Key features
- Mixture-of-experts architecture with 132B total / 36B active parameters
- Open weights for base and instruct variants
- Strong code and reasoning benchmarks at release
- Trained on 12 trillion tokens of curated data
Use cases
- Research on mixture-of-experts model architectures
- Self-hosted LLM deployments on enterprise infrastructure
- Fine-tuning experiments on a large open base model
Pricing
DBRX weights are free to download under the Databricks Open Model License; running the model requires the user's own infrastructure, and Databricks no longer serves it as a hosted model.
Frequently asked questions
What is DBRX?
DBRX is an open-weight mixture-of-experts LLM from Databricks with 132B total parameters, released in March 2024.
Is DBRX still available?
The weights remain available for download, though Databricks retired DBRX from its hosted model serving in 2025.
Can DBRX be used commercially?
Yes, subject to the terms of the Databricks Open Model License that accompanies the weights.
Similar products
- Baichuan — Open-weight bilingual LLM family from Beijing-based Baichuan Intelligence, now focused on medical AI with its Baichuan-M reasoning models.
- Command R — Cohere's open-weight LLM family optimized for retrieval-augmented generation and tool use; succeeded by Command A as Cohere's flagship.
- DeepSeek — Open-weight LLM family and chat assistant from Hangzhou-based DeepSeek, known for strong reasoning models. The official DeepSeek-V4-Flash-0731 API brings a major agent-capability upgrade via post-training, with native Responses API support and Codex adaptation.
- Falcon — Family of open-weight large language models developed by the Technology Innovation Institute (TII).
- Gemma — Google's family of lightweight open-weight models derived from the same research as Gemini.
- GLM — Family of open-weight large language models from Zhipu AI, spanning general, reasoning, and coding-focused variants.
Sources
This record was last reviewed on 2026-07-30.
Machine-readable record: /api/products/dbrx.json · Spot an error? Suggest a correction