June 4th, 2026
Batch Container
Realtime Container
Version 15.9.0 is now available for Batch Container, Real-Time Container, GPU Transcription Inference Container and GPU Translation Inference Container.
HTTP Batch Transcription
Supports URL-based audio input. Users can provide an audio file URL, enabling the service to fetch the file for transcription. See the documentation for more details.
The /ready endpoint now reports engines_used, improving service visibility. See the documentation for more details.
New Arabic models (Enhanced Operating Point) give up to 2.7% relative accuracy improvement.
Fixed transcription job failures when speaker diarization was enabled for WAV files with missing or incorrect duration metadata.
Fixed failure to generate transcripts for a small number of stereo files when one channel contains leading silence.
Software Bill of Materials (SBOM) is available for download from the corresponding release page in our Support Portal.
libgstreamer (CVE-2025-3887): This component has been manually patched to address CVE-2025-3887. Note: Security scanners may still flag this component as vulnerable because the base version string remains unchanged
tritonserver 2.61.0 (NGC 25.09): This is a component in sm-gpu-inference-server and sm-translation-server images that is not listed in the corresponding SBOMs due to lack of installation metadata in the base images published by Nvidia.
The identified CVEs have been reviewed and are not considered exploitable, provided the Triton API is only accessible to transcriber containers within the same trust domain. This can be achieved by deploying STT in a private Kubernetes cluster or by using a service mesh to restrict access to the appropriate pods.
A future release will rebuild against Nvidia NGC 26.03 or later with the relevant fixes.
May 14th, 2026
Realtime Kubernetes
Updates Redis dependency to use 8.6.3-alpine image.
Redis is deployed using community image from a dependent helm chart (“sm-redis”) instead of Bitnami image and helm chart.
Updates Realtime container version to 15.7.0. For more information, see the 15.7.0 container release notes.
Updates sessiongroups CustomResourceDefinitions and controller. Refer to the sm-realtime Helm chart RELEASE_NOTES.md for upgrade details.
Software Bill of Materials (SBOM) is available for download from the corresponding release page in our Support Portal.
libgstreamer (CVE-2025-3887): This component has been manually patched to address CVE-2025-3887. Note: Security scanners may still flag this component as vulnerable because the base version string remains unchanged
The following vulnerabilities were reviewed and determined to have no security impact on this release due to specific configurations or the use of closed, trusted environments:
May 12th, 2026
Batch SaaS
Fixed transcription job failures when speaker diarization was enabled for WAV files with missing or incorrect duration metadata.
Bug introduced in release version: 2026.04.23
Updated Orchestrator Version: 2026.05.08+3477f55380+15.9.0
May 7th, 2026
Batch SaaS
Fixed failure to generate transcripts for a small number of stereo files when one channel contains leading silence.
Resolves the following tickets: 32372
Bug introduced in release version: 2026.04.23
Updated Orchestrator Version: 2026.05.05+c019d07f67+15.8.0
May 1st, 2026
Batch Container
Realtime Container
GPU Transcription Inference Container
GPU Translation Inference Container
Version 15.7.0 is now available for Batch Container, Real-Time Container, GPU Transcription Inference Container and GPU Translation Inference Container.
HTTP Batch Transcription – Processes multiple jobs using persistent workers, reducing turnaround time and improving CPU/GPU utilization. See the documentation for more details.
New model (Enhanced Operating Point) for English (en) improves accuracy across:
Numbers, spellouts, and other alphanumerics
Medical measurements and terminology
Mixed spoken alphanumeric sequences of numbers and characters
Character sequences, for example spell outs of names or abbreviations
Formatting consistency for letter sequences, now returned as upper case letters
Email addresses and web URLs
Large and compound monetary amounts
Improved accuracy at low latency when using ForceEndOfUtterance
Updated Japanese (ja) model for Enhanced Operating Point
Software Bill of Materials (SBOM) is available for download from the corresponding release page in our Support Portal.
libgstreamer (CVE-2025-3887): This component has been manually patched to address CVE-2025-3887. Note: Security scanners may still flag this component as vulnerable because the base version string remains unchanged
April 24th, 2026
Realtime SaaS
New model (Enhanced Operating Point) for English (en) improves accuracy across:
Medical measurements and terminology
Mixed spoken alphanumeric sequences of numbers and characters
Character sequences, for example spell outs of names or abbreviations
Formatting consistency for letter sequences, now returned as upper case letters
Email addresses and web URLs
Large and compound monetary amounts
Improved accuracy at low latency when using ForceEndOfUtterance
Updated Japanese (ja) model for Enhanced Operating Point
Updated Orchestrator Version: 2026.04.21+fd908134bc+15.7.0
April 16th, 2026
Batch SaaS
New model (Enhanced Operating Point) for English (en) improves accuracy across:
Medical measurements and terminology
Mixed spoken alphanumeric sequences of numbers and characters
Character sequences, for example spell outs of names or abbreviations
Formatting consistency for letter sequences, now returned as upper case letters
Email addresses and web URLs
Large and compound monetary amounts
Updated Japanese (ja) model for Enhanced Operating Point
Updated Orchestrator Version: 2026.04.10+192b655fa8+15.5.0
March 12th, 2026
Realtime SaaS
New model (Enhanced Operating Point) for English (en) improves accuracy across numbers, spellouts, and other alphanumerics.
Updated Orchestrator Version: 2026.02.27+2ce3ed4fc8+15.2.0
March 9th, 2026
Batch SaaS
Faster processing for audio files up to 12 minutes when using a custom dictionary.
Supported languages: Arabic, Catalan, Dutch, English, French, German, Greek, Hebrew, Hindi, Japanese, Norwegian, Persian, Portuguese, Russian, Spanish, and Swedish.
New model (Enhanced Operating Point) for English (en) improves accuracy across numbers, spellouts, and other alphanumerics.
Updated Orchestrator Version: 2026.02.27+2ce3ed4fc8+15.2.0
March 5th, 2026
Realtime Kubernetes
Enables repopulation to automatically restore cluster state in the event of Redis data loss.
Supports language and operating-point based model cost generation for the custom inference server recipe. Refer to the sm-realtime Helm chart README.md for configuration details.
Updates Realtime container version to 15.0.0. For more information, see the 15.0.0 container release notes.
Software Bill of Materials (SBOM) is available for download from the corresponding release page in our Support Portal.
libgstreamer (CVE-2025-3887): This component has been manually patched to address CVE-2025-3887. Note: Security scanners may still flag this component as vulnerable because the base version string remains unchanged
Redis versioning: This release by default uses Redis version 8.2.1 with publicly disclosed security vulnerabilities (CVEs) that have been assessed to not impact this product. Users who require additional security controls or remediation for these CVEs can choose to deploy a different Redis version which remediates issues of concern.
The following vulnerabilities (including those reported for the third-party package Redis version) were reviewed and determined to have no security impact on this release due to specific configurations or the use of closed, trusted environments: