[BOT ISSUE] Databricks: add missing databricks-gpt-5-5-pro, databricks-gemini-3-5-flash, databricks-gpt-5-4-mini

[github-actions]

Summary

Three additional high-impact Databricks Foundation Model API pay-per-token models are missing from packages/proxy/schema/model_list.json. This is a follow-up to #559, which covers the top-3 missing models (databricks-claude-opus-4-7, databricks-claude-sonnet-4-6, databricks-gpt-5-5).

The local catalog has only 4 active Databricks-prefixed entries, while the official Databricks docs list 30+ pay-per-token serverless models.

Missing Model IDs

Model ID	Context Window	Max Output	Supported Inputs	Underlying Model
databricks-gpt-5-5-pro	400,000	128,000	text, image	GPT-5.5 Pro
databricks-gemini-3-5-flash	not published	not published	text, image, video, audio	Gemini 3.5 Flash
databricks-gpt-5-4-mini	400,000	128,000	text, image	GPT-5.4 mini

Verification checklist

• Cross-source confirmation: All three model IDs confirmed on two independent Databricks documentation pages:
  1. Foundation model overview — lists all pay-per-token serverless models in the regional availability table
  2. Supported models (Foundation Model APIs) — detailed table with model IDs, context windows, and supported input types
• Recent commits check: No databricks-gpt-5-5-pro, databricks-gemini-3-5-flash, or databricks-gpt-5-4-mini entries exist in model_list.json.
• ID format validation: Follows the existing Databricks naming convention in the catalog: databricks-<model-description> (e.g., databricks-meta-llama-3-3-70b-instruct at line 9156, databricks-meta-llama-3-1-405b-instruct at line 9168).

Verification notes

Field	Source
Model IDs	Foundation model overview — pay-per-token model table
Context windows (400K for GPT-5.5 Pro and GPT-5.4 mini)	Supported models page — context window column
Max output (128K for GPT-5.5 Pro and GPT-5.4 mini)	Supported models page — max output column
Supported input types	Supported models page — supported inputs column
Models are pay-per-token serverless	Foundation model overview — listed under pay-per-token endpoints

Fields NOT verified from official sources:

• input_cost_per_mil_tokens / output_cost_per_mil_tokens — Databricks does not publish stable public per-model pricing on their docs. Skipped per provider-specific constraints.
• max_input_tokens / max_output_tokens for databricks-gemini-3-5-flash — not published on the supported models page.
• multimodal — the underlying models support multimodal input, and the supported models page lists image/video/audio inputs, but Databricks may restrict some modalities.
• reasoning / reasoning_budget — not confirmed for the Databricks-hosted versions specifically.

Local files inspected

• packages/proxy/schema/model_list.json — grep for databricks returns only 5 entries:
  • databricks-claude-3-7-sonnet (deprecated since 2026-02-19)
  • databricks-meta-llama-3-3-70b-instruct
  • databricks-meta-llama-3-1-405b-instruct
  • databricks-meta-llama-3-1-8b-instruct
  • (plus databricks/dbrx-instruct which is an older entry without a provider)

Source URLs

• https://docs.databricks.com/aws/en/machine-learning/model-serving/foundation-model-overview
• https://docs.databricks.com/aws/en/machine-learning/foundation-model-apis/supported-models

{
  "kind": "missing_model",
  "provider": "databricks",
  "models": ["databricks-gpt-5-5-pro", "databricks-gemini-3-5-flash", "databricks-gpt-5-4-mini"],
  "status": "active",
  "model_specs": {
    "databricks-gpt-5-5-pro": {
      "format": "openai",
      "flavor": "chat",
      "multimodal": true,
      "displayName": "GPT-5.5 Pro",
      "max_input_tokens": 400000,
      "max_output_tokens": 128000,
      "available_providers": ["databricks"]
    },
    "databricks-gemini-3-5-flash": {
      "format": "openai",
      "flavor": "chat",
      "multimodal": true,
      "displayName": "Gemini 3.5 Flash",
      "available_providers": ["databricks"]
    },
    "databricks-gpt-5-4-mini": {
      "format": "openai",
      "flavor": "chat",
      "multimodal": true,
      "displayName": "GPT-5.4 mini",
      "max_input_tokens": 400000,
      "max_output_tokens": 128000,
      "available_providers": ["databricks"]
    }
  },
  "source_urls": [
    "https://docs.databricks.com/aws/en/machine-learning/model-serving/foundation-model-overview",
    "https://docs.databricks.com/aws/en/machine-learning/foundation-model-apis/supported-models"
  ]
}

[github-actions]

Autofix could not safely resolve this ticket.

Autofix skipped because the issue does not yet contain enough verified metadata to safely apply catalog changes for databricks-gpt-5-5-pro, databricks-gemini-3-5-flash, databricks-gpt-5-4-mini. Refusing to add databricks-gemini-3-5-flash without verified token limits in the issue metadata

Verified information from this run:

• Provider: databricks
• Parsed models: databricks-gpt-5-5-pro, databricks-gemini-3-5-flash, databricks-gpt-5-4-mini
• Official source URLs already on the ticket: https://docs.databricks.com/aws/en/machine-learning/model-serving/foundation-model-overview, https://docs.databricks.com/aws/en/machine-learning/foundation-model-apis/supported-models

To make autofix work on a future run, update the machine-readable metadata block for this missing model issue and include the verified fields needed for the catalog change, such as:

• model_spec for single-model issues or model_specs for multi-model issues
• displayName
• parent for non-base dated snapshots or location-scoped variants when official sources verify the relationship
• pricing fields
• max_input_tokens
• max_output_tokens
• available_providers
• deprecated / deprecation_date
• locations when the provider/model requires explicit location metadata