We are pleased to announce the addition of mistral-small-3 and the introduction of llama-3-3-70b-turbo, which replaces llama-3-3-70b-it for better cost efficiency.
mistral-small-3 model is also available in Openrouter from Ubicloud provider. For more information, visit our Openrouter profile
Our AI inference endpoints stack has been integrated with FlashInfer, a high-performance library and kernel generator designed for LLMs, to provide higher inference performance. Learn more about this enhancement in our GitHub pull request here.
To streamline our AI inference endpoints collection, we have removed some lesser-used models, simplifying our model catalog.PR #2824, #2869, #2840, #2816, #2818
Built-in connection pooling available on all plans
Ubicloud Postgres now comes with PgBouncer preinstalled. By connecting to PgBouncer at port 6432, you can scale your connections to Postgres to higher levels.
With the newest 0.15.6 version, ParadeDB on Ubicloud Postgres gains the paradedb.match query function, JSON support in paradedb.term_set, along with stability improvements and performance enhancements.https://docs.paradedb.com/changelog/0.15.6PR #2818
We have developed a CLI tool, providing a secure and programmatic way to interact with your Ubicloud resources. The tool is available for preview in the Ubicloud repository.PR #2671, #2857