Databricks—Databricks released DBRX as a fully open-weight 132B-parameter mixture-of-experts foundation model
On March 27, 2024 Databricks released DBRX, a 132-billion-parameter mixture-of-experts large language model, under a permissive open-weight license including commercial use rights. At release DBRX outperformed Llama 2 70B and Mixtral 8x7B on standard benchmarks and was among the largest fully open-weight models of its time. Databricks published the training methodology, evaluation data and tokenizer alongside the weights, contrasting with closed releases from OpenAI, Anthropic and Google.
Scoring Impact
| Topic | Direction | Relevance | Contribution |
|---|---|---|---|
| AI Oversight | +toward | primary | +1.00 |
| Knowledge Access & Information Freedom | +toward | secondary | +0.50 |
| Open Source | +toward | primary | +1.00 |
| Overall incident score = | +0.715 | ||
Score = avg(topic contributions) × significance (high ×1.5) × confidence (0.57)
Evidence (1 signal)
Databricks released DBRX-Base and DBRX-Instruct under permissive open-weight license
Databricks' March 27, 2024 announcement and accompanying technical report released the 132B-parameter DBRX mixture-of-experts model under permissive open-weight license including commercial use, with training methodology and tokenizer disclosure.