Changelog
See our latest feature releases, product improvements and bug fixes
View detailed billing and usage metrics
You can now view a daily breakdown of your model usage and billing information to get more insight into usage and costs. Here are the key changes: A new graph displays daily costs, requests, and...
Feb 6, 2024Double inference speed and throughput with NVIDIA H100 GPUs
Baseten is now offering model inference on H100 GPUs starting at $9.984/hour. Switching to H100s offers a 18 to 45 percent improvement in price to performance vs equivalent A100 workloads using...
Jan 19, 2024Deploy state-of-the-art open source models instantly
We’ve totally refreshed our model library to make it easier for you to find, evaluate, deploy, and build on state-of-the-art open source ML models. You can try the new model library for yourself...
Jan 11, 2024NVIDIA L4 GPUs now generally available on Baseten
You can now deploy models to instances powered by the L4 GPU on Baseten. NVIDIA’s L4 GPU is an Ada Lovelace series GPU with: 121 teraFLOPS of float16 compute 24 GB of VRAM at a 300 GB/s memory...
Jan 8, 2024Give names to model deployments
When deploying with Truss via truss push , you can now assign meaningful names to your deployments using the --deployment-name argument, making them easier to identify and manage. Here's an example:...
Dec 15, 2023Updated defaults and language for autoscaling settings
Autoscaling lets your deployed models handle variable traffic while making efficient use of model resources. We’ve updated some language and default settings to make using autoscaling more intuitive....
Nov 10, 2023Retry failed builds and deploys
You can now retry failed model builds and deploys directly from the model dashboard in your Baseten workspace. Model builds and deploys can fail due to temporary issues, like a network error while...
Oct 31, 2023Overhauled model management experience
We've made some big changes to the model management experience to clarify the model lifecycle and better follow concepts you're already familiar with as a developer. These changes aren't breaking...
Oct 27, 2023Add workspace API keys for more granular permissions
We added workspace API keys to give you more control over how you call models, especially in production environments. There are now two types of API keys on Baseten: Personal keys are tied to your...
Oct 16, 2023New model IDs for deployed models
The model IDs for some models deployed on Baseten have been changed. This is not a breaking change. All existing model invocations using the old model IDs will continue to be supported. You do not...