Sauce Labs AURA — Model Choice is an enterprise AI testing feature announced by Sauce Labs on August 19, 2026. It enables organizations to select the Large Language Model (LLM) used within AURA’s automated workflow, which spans test creation and execution, failure diagnosis, and self-healing. The core principle is a Bring Your Own Model (BYOM) approach that allows enterprises to apply trusted options from open-source, open-weight, or commercial models rather than locking the testing platform to a single specific model. Just as GPT can swap multiple models for a single text task, Model Choice decouples model selection and test orchestration in the software verification process, providing flexibility to adapt to technological changes. While existing AI testing tools can rapidly generate tests or explain failures, they may depend on the platform-specified models and data processing methods. If enterprises need to adopt different models due to security policies, regulatory compliance, cost, response quality, or internal model strategies, they often face the burden of reconfiguring test pipelines and governance. The differentiator of Model Choice is its focus on maintaining the same test execution and release verification system even when models are swapped. In other words, models are treated as replaceable engines, while the verification flow that connects execution results and operational errors obtained from real or virtual devices remains within the platform. This enables the construction of a closed-loop validation loop where tests and fixes proposed by AI are re-verified against actual execution results. From a quality engineering perspective, this feature can be utilized in workflows where tests are written based on new feature requirements, executed on real or virtual devices, and failure logs are diagnosed by AURA to identify unstable selectors or changed UI elements as targets for self-healing. By operating the same test suite under different LLM selection policies, organizations can compare test generation quality and failure diagnosis results while maintaining existing release gates. However, since the supported model list, connection methods per model, evaluation metrics, and detailed settings cannot be confirmed solely from the official announcement, it is necessary to verify product documentation or contract terms with Sauce Labs prior to actual adoption. Organizations developing life sciences software can apply this feature to products where change verification and traceability are critical, such as laboratory equipment control apps, mobile interfaces for clinical research, and research data portals. For example, regression tests can be generated from requirements, executed on real or virtual devices, and failure causes diagnosed to link to release approval documentation. The ability to maintain the same verification procedure while selecting approved models according to regulatory or security policies may reduce the burden of reconstructing test systems with every model change. However, features that automatically ensure compliance with medical device software regulations or clinical validation were not confirmed in the input materials, and final approval requires independent review and verification by the organization.

Sauce Labs AURA — Model Choice is an enterprise AI testing feature announced by Sauce Labs on August 19, 2026. It enables organizations to select the Large Language Model (LLM) they wish to use within AURA’s automated workflow, which spans test creation and execution, failure diagnosis, and self-healing. The core principle is a Bring Your Own Model (BYOM) approach that allows companies to apply their trusted choices among open-source, open-weight, and commercial models rather than locking the testing platform to a single specific model.

🔬 Regression Verification of Laboratory Equipment Control App: Write tests based on requirements, execute them on physical and virtual devices, and apply failure diagnosis and self-healing. This scenario restricts model selection policies using an approved LLM and links execution results to release reviews.