February 12th, 2026
Batch SaaS
Updated models for Swedish (sv) with improved transcription accuracy (reduction in incorrect insertions)
“x-ray” and “x-rays” are now always hyphenated in English (en) transcription output
Updated Orchestrator Version: 2026.02.06+686b83b434+15.0.0
February 5th, 2026
Real-Time Kubernetes
Adds support for using Realtime components without a ClusterRole, disabling cluster-wide access for the SessionGroups controller. Refer to the sm-realtime Helm chart README.md for configuration details.
This Helm chart version uses Realtime container version 14.13.0. For more information, see the 14.13.0 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:
February 3rd, 2026
Batch SaaS
Significantly faster processing for Arabic, Catalan, Dutch, French, German, Greek, Hebrew, Hindi, Japanese, Norwegian, Persian, Portuguese, Russian and Swedish audio files up to 12 minutes in duration - released to all endpoints
January 26th, 2026
Batch Container
Real-Time Container
GPU Transcription Inference Container
GPU Translation Inference Container
Version 14.13.0 is now available for Batch Container, Real-Time Container, GPU Transcription Inference Container and GPU Translation Inference Container.
New Multilingual transcription language pack Mandarin Malay Tamil English (cmn_en_ms_ta) available now. Refer to the documentation for more details.
New medical domain-specific models for Danish, Norwegian, Swedish giving the highest accuracy for healthcare use cases. Refer to the documentation for more details.
New models for Bilingual Malay English, Bilingual Tamil English, Danish, Finnish, Maltese, Norwegian, Swedish with improved transcription accuracy
New models for Bilingual Malay English, Bilingual Tamil English, Danish, Finnish, Norwegian, Swedish with improved transcription accuracy
New models for Danish, Finnish, Galician, Norwegian, Swedish with improved transcription accuracy
New models for Danish, Finnish, Galician, Maltese, Norwegian, Swedish with improved transcription accuracy
Fix for number formatting in Malay & English bilingual (en_ms) and Tamil & English bilingual (en_ta) language packs
Resolves the following tickets: 30055
Fix for incorrect casing in Japanese (ja) transcription output
Resolves the following ticket: 28276
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 for this release.
January 21st, 2026
Batch SaaS
Real-Time SaaS
Fix for number formatting in Mandarin Malay Tamil & English multilingual (cmn_en_ms_ta), Malay & English bilingual (en_ms) and Tamil & English bilingual (en_ta) language packs.
Resolves the following tickets: 30452, 30055
Fix for incorrect casing in Japanese (ja) transcription output.
Resolves the following ticket: 28276
Updated Orchestrator Version: 2026.01.15+504a3b4d7c+14.13.0
January 15th, 2026
Real-Time SaaS
Medical domain-specific models (Enhanced Operating Point) for Swedish (sv) giving the highest accuracy for healthcare use cases. Refer to the documentation for more details.
Relative accuracy improvements (medical domain): 60%
New models for Swedish (sv) with improved transcription accuracy
Relative accuracy improvements (general domain): Enhanced Operating Point - 9.3%; Standard Operating Point - 1.8%
New models for Norwegian (no) with improved transcription accuracy for general and medical domains
Relative accuracy improvements (general domain): Enhanced Operating Point - 6.8%; Standard Operating Point - 4.7%
Relative accuracy improvements (medical domain, Enhanced Operating Point): 11%
New models for Finnish (fi) with improved transcription accuracy for general and medical domains
Relative accuracy improvements (general domain): Enhanced Operating Point - 4.9%; Standard Operating Point - 14.1%
Relative accuracy improvements (medical domain, Enhanced Operating Point): 10.4%
Updated models for Galician (gl) and Maltese (mt) for Enhanced Operating Point
Custom Dictionary now only allows entries with words up to 4000 characters. Any entries exceeding this limit will be automatically removed before transcription begins. Refer to documentation here for details.
Updated Orchestrator Version: 2026.01.09+e449221ca0+14.12.0
January 14th, 2026
Batch SaaS
Medical domain-specific models (Enhanced Operating Point) for Swedish (sv) giving the highest accuracy for healthcare use cases. Refer to the documentation for more details.
Relative accuracy improvements (medical domain): 60%
New models for Swedish (sv) with improved transcription accuracy
Relative accuracy improvements (general domain): Enhanced Operating Point - 9.3%; Standard Operating Point - 1.8%
New models for Norwegian (no) with improved transcription accuracy for general and medical domains
Relative accuracy improvements (general domain): Enhanced Operating Point - 6.8%; Standard Operating Point - 4.7%
Relative accuracy improvements (medical domain, Enhanced Operating Point): 11%
New models for Finnish (fi) with improved transcription accuracy for general and medical domains
Relative accuracy improvements (general domain): Enhanced Operating Point - 4.9%; Standard Operating Point - 14.1%
Relative accuracy improvements (medical domain, Enhanced Operating Point): 10.4%
Updated models for Galician (gl) and Maltese (mt) for Enhanced Operating Point
Custom Dictionary now only allows entries with words up to 4000 characters. Any entries exceeding this limit will be automatically removed before transcription begins. Refer to documentation here for details.
Updated Orchestrator Version: 2026.01.09+e449221ca0+14.12.0
January 8th, 2026
Real-Time Kubernetes
Introducing the first release of Helm charts for deploying and managing real-time applications on Kubernetes. Please refer to the documentation for deployment guides.
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). 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:
December 24th, 2025
Batch SaaS
Fix for an issue where Speaker identification failed to label speakers in some jobs
Resolves the following ticket: 30188
December 16th, 2025
Real-Time SaaS
Faster finals with force end of utterance – Enables the client to force finalise transcription at the end of speech to produce faster finals (~200ms), ideal when using external VAD or turn detection models for voice agents. Refer to documentation here for more details.