Releasesby GlyphThis Release Contains The Following Changes
- Flux Word-Level Timestamps , each word in the words array now includes start and end times on the Flux streaming endpoint ( /v2/listen ). This is an additive change. If your client library rejects unrecognized response fields, update to a current SDK version , for the Deepgram Python SDK, use 6.0.1 or later, or 5.3.4 if you are on the 5.x line.
Also shipped Jun 30, 2026Deepgram in June 2026 →
Analytical search (Public Preview)
Analytical search, an orchestration capability in Cortex Agents that enables analytical queries over large document collections, is now in Public Preview. Traditional retrieval-augmented generation (RAG) approaches retrieve a small number of passages, which limits their ability to answer questions requiring comprehensive analysis, such as counts...
MATCH_RECOGNIZE is now in Beta
Databricks SQL and Databricks Runtime now support the SQL:2016 MATCH_RECOGNIZE clause for row pattern recognition. Enable the Match Recognize preview in workspace preview settings . See MATCH_RECOGNIZE clause .
Lakehouse Real-Time (Beta)
Lakehouse Real-Time (Lakehouse//RT) is now in Beta . Lakehouse//RT is a serverless SQL warehouse type built for low-latency, high-concurrency workloads, delivering sub-second SQL read queries against Unity Catalog tables.
Gemini Omni Flash in public preview
Released gemini-omni-flash-preview, a high-performance multimodal model designed for high-speed video generation and conversational video editing. Using the Interactions API, you can generate 3--10 second videos at 720p from text descriptions or animate still images, and then conversationally edit and refine the outputs.
Gemini 3.1 Flash Lite Image (Nano Banana 2 Lite) is generally available
gemini-3.1-flash-lite-image is a multimodal model optimized for ultra-low latency, cost-effective image generation and editing, now GA in the Gemini API.
[Launched] Generally Available: Toolboxes in Microsoft Foundry
Microsoft Foundry brings Toolboxes to GA, providing a unified endpoint to invoke the most-used tools in prompt agents and callable outside of prompt agents. Today every agent team hand-assembles its own tool list to solve the same tasks, with no shared ar
More from DeepgramEvery Deepgram release →
Container Images (release 260630)
- quay.io/deepgram/self-hosted-api:release-260630 Equivalent image to: quay.io/deepgram/self-hosted-api:1.192.3-1 quay.io/deepgram/self-hosted-engine:release-260630 Equivalent image to: quay.io/deepgram/self-hosted-engine:3.119.1-1 Minimum required NVIDIA driver version: >=570.172.08 quay.io/deepgram/self-hosted-license-proxy:release-260630 Equivalent image...
Breaking Change: GET /v1/models Response Format
The self-hosted GET /v1/models endpoint now lists every loaded model, not just core speech-to-text (Nova) models. The response now also includes Text-to-Speech (Aura voice packs and voices) and helper models (entity detection/NER, diarization), groups results by model type, and indicates batch versus streaming mode for each speech-to-text model.
Claude Sonnet 5 Now Available
claude-sonnet-5 is now available as a managed Anthropic LLM in the Voice Agent API. This Advanced tier model delivers improved reasoning and conversational quality for your voice agents. Set the model in your agent configuration: { "agent": { "think": { "provider": { "type": "anthropic", "model": "claude-sonnet-5", "temperature": 0.5 } } } }
Deprecation: Claude Sonnet 4
claude-sonnet-4-20250514 is now deprecated. Migrate to claude-sonnet-5 or another supported Anthropic model. This model has been retired by Anthropic and will produce an error if specified in your settings configuration. For the full list of supported models and pricing tiers, see the Voice Agent LLM Models documentation .
🆕 Word-Level Timestamps Now Available in Flux
Flux now includes word-level timestamps in its responses. Each word in the words array carries start and end times (type double ) alongside the existing word and confidence values.
SDK Compatibility
Most integrations will continue to work without changes. The exception is clients configured to reject unrecognized fields, which may return errors when the new fields appear. Here is what we recommend: Python : Versions 6.0.1 and later are unaffected. If you are on the 5.x line, update to 5.3.4.
🆕 Redaction: Expanded Language Coverage & Flux Support
- Number redaction ( numbers , true , aggressive_numbers ) now supports these languages on the Hosted API: Danish, Dutch, English, French, German, German (Swiss), Italian, Norwegian, Polish, Portuguese, Spanish, and Swedish. On multilingual models, number redaction applies the correct rules to each detected language within a transcript.
InjectAgentMessage: behavior field and new interrupt mode
The InjectAgentMessage message now accepts a behavior field that controls how an injected message interacts with any in-progress user or agent turn. The default and queue behaviors were previously documented; the new interrupt behavior is introduced with this release. default , the agent speaks only if neither the user nor the agent is mid-turn.