Hi team, we're running into issues with our Knowledge Bases and wanted to flag this. Context: The KB model dropdown currently lists models from the Gemini 2.0 and earlier families. Google has been deprecating these: -

gemini-2.0-flash-001

and

gemini-2.0-flash-lite-001

are only available to existing customers as of March 6, 2026, and are scheduled for full shutdown on June 1, 2026 (source: https://docs.cloud.google.com/vertex-ai/generative-ai/docs/models/gemini/2-0-flash) - All Gemini 1.0 and 1.5 models have already been shut down by Google (source: https://ai.google.dev/gemini-api/docs/deprecations) What we're seeing: - With a new Google API key:

404 — "This model models/gemini-2.0-flash is no longer available to new users"

- With an existing Google API key:

429 — "Resource exhausted"

— however, our Google AI Studio dashboard shows we're well within our rate limits (3/2K RPM, 61K/4M TPM, 13/Unlimited RPD). This appears to be either a known Google-side issue with deprecated models (there are widespread reports of "ghost 429s" on deprecated Gemini models in early 2026), or something on Vapi's infrastructure side. Either way, it's not something we can resolve from our end. The ask: Could the team look into adding

gemini-2.5-flash

and

gemini-2.5-flash-lite

as options in the Knowledge Base model selector? These are Google's recommended replacements. Right now the KB feature is difficult to use since all the listed models are either deprecated or being phased out. Happy to provide call IDs, screenshots, or anything else that helps. This is blocking our deployment so any update on timeline would be appreciated. A Call ID for reference - c8f5c652-6ce6-4640-9552-27e7d5e75174 This is affecting the uptime and I would request any immediate support or solution if possible. https://cdn.discordapp.com/attachments/1496069386136719481/1496069386316943390/kb_model_list.jpeg?ex=69e88ab1&is=69e73931&hm=607120651b85166172b072eac5d6f2b7ffcb747d86af6be8a05eae23c508a94b&

Actionable Suggestion: Currently, the Knowledge Base model selector only supports Gemini 2.0 and earlier models, which are being deprecated. While the new

gemini-2.5-flash-preview-05-20

model is supported in Vapi, it is not yet confirmed for Knowledge Base integration. To prepare for when new models become available, you can configure your Knowledge Base tool via API as shown below—just update the

"model"

field when

gemini-2.5-flash

is officially supported:

json
{
  "type": "query",
  "function": { "name": "product-query" },
  "knowledgeBases": [
    {
      "provider": "google",
      "name": "product-kb",
      "model": "gemini-2.5-flash", // Use this when available
      "description": "Contains comprehensive product information",
      "fileIds": ["your-file-id-1", "your-file-id-2"]
    }
  ]
}

Monitor the [Vapi changelog](https://docs.vapi.ai/changelog/2025/5/24) for updates on Gemini 2.5 model support in Knowledge Bases. Sources: - [March 9, 2025 Changelog](https://docs.vapi.ai/changelog/2025/3/9) - [May 24, 2025 Changelog](https://docs.vapi.ai/changelog/2025/5/24) - [Using the Query Tool for Knowledge Bases](https://docs.vapi.ai/knowledge-base/using-query-tool)