Artificial intelligence news.
Straight from the original announcements.

AWS brings OpenTSLM time-series model to SageMaker JumpStart

November 17, 2026, 19:06By the AI News Bites Editorial TeamSource: AWS
AWS brings OpenTSLM time-series model to SageMaker JumpStart
Image: AWS

AWS has made OpenTSLM, Aionic Labs’ open source family of time-series language models, available through Amazon SageMaker JumpStart. The deployment enables organizations to run the model in their own AWS environments and combine operational measurements with text context.

Key points
  • OpenTSLM is now available in Amazon SageMaker JumpStart.
  • OpenTSLM combines time-series measurements with text context.
  • OpenTSLM-Flamingo has 5.24 billion parameters and uses Gemma 3 4B.
  • Nexthink and Aionic Labs are applying the technology to proactive IT support.
  • AWS provides a tutorial for fine-tuning OpenTSLM with TimeNet and ARFBench data.

A model for operational data and context

In its announcement, AWS said OpenTSLM is designed to help organizations interpret measurements such as equipment sensor data, supply and demand, and device telemetry. The models can process those signals alongside maintenance records, market news and engineers’ observations, helping systems identify emerging problems in recent measurements.

AWS said general-purpose models such as Claude or ChatGPT can analyze operational data, but converting long histories across hundreds of metrics into text or images can consume large numbers of tokens. OpenTSLM is intended to work more directly with time-series information, so experts can ask follow-up questions about what changed and how to respond.

TimeNet, described by AWS as shared data infrastructure for time-series data, provides a format and API for finding, building and loading datasets used to train and evaluate the models.

Nexthink uses the approach for proactive IT support

AWS highlighted work by Nexthink and Aionic Labs on Spark, Nexthink’s personal IT agent. Spark combines device telemetry with contextual information to identify issues employees may experience, guide them through a possible fix and follow up on the result.

The workflow can use CPU activity, memory use, available disk space, application stability and connectivity, together with timestamped events and device and software context. According to AWS, OpenTSLM provides Spark with three outputs: a state summary that explains the likely issue and supporting signal patterns, a recommended action, and an expected outcome.

In AWS’s illustrative low-disk example, the model connects declining free space, a rising disk queue and more frequent application hangs with symptoms such as a slow computer or difficulty saving files. Spark checks the available space and whether temporary files can be removed through an IT-approved cleanup. It asks the employee for confirmation before running the cleanup, checks the result and offers to pass the findings to IT support if the issue remains.

OpenTSLM-Flamingo is available through SageMaker

The model offered in SageMaker JumpStart is OpenTSLM-Flamingo. AWS said it combines the Gemma 3 4B language model with a time-series encoder and cross-attention layers, allowing the language model to attend to encoded series. The model has 5.24 billion parameters stored in bfloat16, with weights requiring about 10.5 GB of GPU memory.

SageMaker JumpStart provides model discovery and deployment, while SageMaker AI supports inference endpoints within an organization’s AWS infrastructure. AWS lists pre-built GPU-accelerated inference containers, automatic scaling endpoints, standardized inference APIs, integrated monitoring and enterprise security options among the platform’s capabilities.

OpenTSLM runs on a single GPU with at least 24 GB of memory and bfloat16 support. AWS recommends 32 GB of host memory, and gives ml.g6.2xlarge, with one NVIDIA L4 GPU and 24 GB of memory, as the default instance type. The announcement also describes a tutorial for fine-tuning an OpenTSLM checkpoint from Hugging Face with ARFBench data loaded through TimeNet, storing the resulting model artifact in Amazon S3 before evaluation on a separate SageMaker real-time inference endpoint.

For users, the release provides a path to test time-series reasoning in applications that combine numerical data and language context. AWS said the same approach can support manufacturing, energy and connected products, while applications can control the observation window, select relevant signals and validate model responses before using them in another workflow.

Related news

AWS brings PolyAI’s Dialog-RSN-1 voice model to SageMaker JumpStart
ModelsNovember 19, 2026, 18:28

AWS brings PolyAI’s Dialog-RSN-1 voice model to SageMaker JumpStart

PolyAI’s Dialog-RSN-1 reasoning model for real-time customer conversations is now available through AWS Marketplace and Amazon SageMaker JumpStart. AWS customers can…

AWS Launches SageMaker JumpStart Leaderboard for Model Comparisons
Products and toolsNovember 18, 2026, 23:31

AWS Launches SageMaker JumpStart Leaderboard for Model Comparisons

AWS has launched the Amazon SageMaker JumpStart Model Leaderboard, a SageMaker Studio view that compares open-weight models on quality, performance and estimated cost.…

Cloudflare releases open-weight Clairic models for document extraction
ModelsNovember 18, 2026, 23:00

Cloudflare releases open-weight Clairic models for document extraction

Cloudflare has released updated Clairic and Clairic-flash, two open-weight models designed to extract data from text, images and PDFs. The models are hosted on Workers…