Expanded LLM Support and More in Releases 2026.17/18

• claude-sonnet-5 – Anthropic's balanced model for day-to-day agent and conversational tasks.
• claude-opus-4-7 & claude-opus-4-8 – Anthropic's most capable models, built for complex reasoning and multi-step workflows.
• gemini-3.6-flash & gemini-3.5-flash-lite – Faster, lower-cost Gemini options for high-volume Generative AI tasks.
• gemini-3.7-flash – Reliably follows instructions and handles multi-step planning with tool-calling support.

The platform also improves general compatibility with Anthropic models by automatically removing sampling parameters that a given model doesn't support.

Other Improvements

Cognigy

• Moved the Logs API to its own service. The endpoint routes remain the same as those used by the Cognigy API, so no migration is required
• Removed the OAuth2 authentication option as a supported API authentication method from the Cognigy API
• Added automatic login via SSO when opening a Cognigy URL in CXone-integrated installations
• Added the Use deployment name as model toggle to the Azure OpenAI LLM provider settings for the Responses API. You can now choose whether to use the Deployment Name or Model parameter to specify the model
• Added Custom Model Options and Custom Request Options to the LLM Entity Extract and Question Nodes. You can now pass additional model and request parameters directly to the LLM call
• Added support for multiple Microsoft SSO OAuth connections in Azure Bot Services Endpoints. You can now manage tokens for each connection separately using the Overwrite Connection option in the Get Token, Invalidate Token, and Token Status Nodes
• Improved compatibility with Anthropic models that don’t support sampling parameters. Now, Cognigy removes unsupported temperature, top_p, and top_k parameters before sending the request
• Improved the reliability of the Project Overview dashboard. If the simulation search fails, the Recently Changed panel still displays other resources instead of showing an error
• Added traceId and sessionId to transformer error logs, making it easier to trace errors to a specific request and session

Voice Gateway

• Added the recordProcessor attribute to the background recording span in the call detail record traces. This attribute identifies whether the audio manager or API server handled the recording