What changed
AWS has added OpenTSLM, an open-source time-series language model (TSLM), to its SageMaker JumpStart service. This model is developed by Aionic Labs and is designed to interpret operational measurements, such as equipment sensor data, supply and demand metrics, and maintenance records, alongside contextual information like market news and engineer observations. OpenTSLM is now available for deployment in AWS environments, enabling organizations to analyze time-series data and identify emerging issues.
What this means in practice
The model helps organizations understand how their systems change over time and take proactive measures before problems escalate. For instance, Nexthink has integrated OpenTSLM into its Spark IT agent to provide proactive support. Spark uses device telemetry and context to identify issues like slow performance, application freezes, or failed file synchronizations. The TSLM provides a state summary, recommended action, and expected outcome to guide employees toward solutions. This approach allows IT teams to address issues before they impact productivity.
Limitations and availability
OpenTSLM is available through Amazon SageMaker JumpStart, which provides a deployment path for the model. The model, named OpenTSLM-Flamingo, combines the Gemma 3 4B language model with a time-series encoder and cross-attention layers. It has a total of 5.24 billion parameters, stored in bfloat16, requiring about 10.5 GB of GPU memory. The model is part of the AWS AI offerings and complements Amazon Bedrock. The supplied announcement does not specify the exact licensing conditions or download versus API availability details.
