DBRX Models by Databricks Mosaic
Details
Capabilities
About
The DBRX family of large language models (LLMs), developed by Databricks, includes DBRX Base and DBRX Instruct, both of which have been released under an open license. These transformer-based, decoder-only models employ a distinguished fine-grained mixture-of-experts (MoE) architecture. This approach leverages a larger number of smaller expert networks, enhancing the quality of output compared to models with fewer, larger experts. Trained on an extensive dataset of 12 trillion tokens encompassing text and code, these models showcase high performance with the capability to handle up to 32,000 tokens of context. DBRX Instruct is specially optimized for few-turn interactions, making it suitable for efficient conversational applications. The DBRX models not only excel in efficiency during training and inference but also outperform many leading closed and open-source models in various benchmarks. They are accessible for download on Hugging Face and integrated through Databricks' Foundation Model APIs.
Current Variants
Use-when guidance is derived from seed capabilities, context, release, and replacement fields.
Use when the workload needs 32k context and 229B parameters.
Use when the workload needs 32k context and 20B parameters.
Use when the workload needs 32k context and 132B parameters.
Use when the workload needs 33k context, 132B parameters, and structured outputs.
Use when the workload needs 33k context and 132B parameters.
Use when the workload needs 32k context and 132B parameters.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| DBRX Expanse | Use when the workload needs 32k context and 229B parameters. | 2024-11 | 32k context229B parameters | Current |
| DBRX Instruct 20B | Use when the workload needs 32k context and 20B parameters. | 2024-07 | 32k context20B parameters | Current |
| DBRX | Use when the workload needs 32k context and 132B parameters. | 2024-03 | 32k context132B parameters | Current |
| DeepInfra DBRX Instruct | Use when the workload needs 33k context, 132B parameters, and structured outputs. | 2024-03 | 33k context132B parametersstructured outputs | Current |
| Fireworks DBRX-Instruct | Use when the workload needs 33k context and 132B parameters. | 2024-03 | 33k context132B parameters | Current |
| DBRX Base | Use when the workload needs 32k context and 132B parameters. | 2024-03 | 32k context132B parameters | Current |
Release Timeline
3 release groupsSpecifications(7 models)
| Model | Released | Context | Parameters | Structured Outputs |
|---|---|---|---|---|
| DBRX Expanse | 2024-11 | 32k | 229B | No |
| DBRX Instruct 20B | 2024-07 | 32k | 20B | No |
| DBRX | 2024-03 | 32k | 132B | No |
| DeepInfra DBRX Instruct | 2024-03 | 33k | 132B | Yes |
| Fireworks DBRX-Instruct | 2024-03 | 33k | 132B | No |
| DBRX Base | 2024-03 | 32k | 132B | No |
Available From(6 providers)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| DeepInfra DBRX Instruct | DeepInfra | $0.6 | $1.2 | Serverless |
| Fireworks DBRX-Instruct | Fireworks AI | $1.5 | $1.5 | Serverless |
Frequently Asked Questions
- What is DBRX used for?
- DBRX is used for structured outputs and coding. The family description and listed model capabilities point to those workloads as the best fit.
- How does DBRX compare to MOSS-Audio?
- DBRX by Databricks Mosaic is strongest where you need structured outputs, while MOSS-Audio by MOSI AI is the closest related family to check for multimodal. DBRX has 7 listed variants and reaches up to 33k context, so compare the specs and pricing tables before choosing a production model.
- Which DBRX model should I use?
- For the lowest listed input price, start with DBRX Instruct through DeepInfra at $0.6/1M input tokens. For the most capable/latest local choice, evaluate DeepInfra DBRX Instruct with 33k context and structured outputs.
