Releasesby Glyph🆕 New languages
- Afrikaans ( af , af-ZA ) , available for batch and streaming. Georgian ( ka , ka-GE ) , available for batch and streaming. Access these languages by setting model="nova-3" and the relevant language code in your request.
Also shipped Aug 17, 2026Deepgram in August 2026 →
AI mode for sensitive data classification (Public Preview)
Sensitive data classification now supports AI mode in public preview. When you enable AI mode on a classification profile, Snowflake uses the OpenAI GPT-5 Mini model to identify additional semantic categories beyond those detected by standard classification, which can improve accuracy.
New SNOWFLAKE_COCO_USAGE_HISTORY view in Account Usage and Organization Usage (General availability)
The SNOWFLAKE_COCO_USAGE_HISTORY view is now generally available in the ACCOUNT_USAGE and ORGANIZATION_USAGE schemas. It provides a unified view of usage history for CoCo across the CLI, Desktop, and Snowsight interfaces. The information in the view includes the number of credits consumed each time a user interacts with CoCo.
Upcoming deprecation of the data quality monitoring downstream_impact column
The downstream_impact column in the data quality monitoring results system table is being deprecated. Deprecation begins September 7, 2026, and rolls out gradually across regions over the following week. Until the rollout reaches your workspace, you might still be able to query downstream_impact .
Upcoming deprecation of the data quality monitoring root_cause_analysis column
The root_cause_analysis column in the data quality monitoring results system table is being deprecated. Deprecation begins August 18, 2026, and rolls out gradually across regions over the following several days. Until the rollout reaches your workspace, you might still be able to query root_cause_analysis .
Session restore for serverless jobs is now in Beta
Restore the Python variables and Spark session from a serverless compute job run into a new interactive notebook, so you can debug a failure or explore results without rerunning the job. Databricks automatically captures a state snapshot when an eligible run fails, runs 30 minutes or longer, or is cancelled, and keeps it for 7 days after the run completes.
Identity attributes in ABAC column mask policies (Beta)
Identity attributes are in Beta. You can now write Unity Catalog ABAC column mask policy conditions that target users by attributes synced from your identity provider, such as department or country , in addition to group membership.
More from DeepgramEvery Deepgram release →
Nova-3 Adds Afrikaans and Georgian, Plus Improved Models for Hungarian, Macedonian, Russian, Slovak, Slovenian, and Urdu
We have added Afrikaans and Georgian as new Nova-3 languages and released improved Nova-3 monolingual models for several existing languages. These updates expand language coverage and enhance transcription quality for batch and streaming workloads.
Improved models
Batch and streaming models Russian ( ru ) Slovak ( sk ) Slovenian ( sl ) Urdu ( ur ) Streaming models Hungarian ( hu ) Macedonian ( mk ) You do not need to change your API requests to access the improvements to existing languages; the updates are live for all users.
Deepgram CLI 0.3.0: Flux TTS and Flux STT
Deepgram CLI 0.3.0 is now available. It completes CLI support for Flux TTS , which reached general availability on 12 August, fixes Flux STT streaming, and enforces the CLI's exit-code contract. Upgrading from 0.2.26 picks up everything in this release. 0.2.27 was tagged but never published, so its changes arrive here.
Full Flux TTS support on dg speak
dg speak now defaults to flux-alexis-en on Speak v2 (WebSocket streaming) instead of Aura 2. Synthesised audio differs from earlier releases unless you request an Aura model explicitly: dg speak "Hello from Deepgram" -m aura-2-asteria-en -o hello.mp3 Two Flux-only controls are now available.
Flux STT streaming is fixed
dg listen has routed flux-* models to Listen v2 since 0.2.5 , but v2 was receiving v1-only parameters ( language , smart_format , punctuate , channels , diarize , interim_results ) and returning HTTP 400.
Redaction and numerals on dg listen
dg listen call.wav --redact numbers --numerals --redact is repeatable. Flux STT accepts numbers and aggressive_numbers ; Listen v1 also accepts categories such as pci and ssn . --numerals converts spoken numbers to digits. Both apply to prerecorded and live paths. All of the above runs on Deepgram Python SDK 7.7.0.
Exit codes are now enforced
This release includes a breaking change. dg previously exited 0 regardless of outcome. It now follows the documented contract: Code Meaning 0 Success 1 Error, including crashes and usage errors such as an unknown command or invalid flag 2 User interrupt: Ctrl-C, or Ctrl-D at a prompt Scripts and CI steps that ignored the exit code will now surface failures...
Error output moved to stderr
Usage errors, crashes, and cancellation messages are written to stderr instead of stdout, so dg -o json keeps stdout machine-readable when a command fails this way. Previously these printed to stdout, which corrupted output piped into tools such as jq . This covers the root command path.