Skip to content

Conversation

@superhighfives
Copy link
Contributor

This pull request updates the Cloudflare AI Gateway provider to improve model metadata consistency, accuracy, and usability. It standardizes model display names, corrects and enriches model configuration files (such as release dates, capabilities, and costs), and reorganizes model files for better structure and discoverability.

Key changes:

1. Model Metadata and Capabilities Updates

  • Standardized and corrected model names, release dates, last updated dates, and feature flags (e.g., attachment, reasoning, tool_call, structured_output, knowledge) across all OpenAI model TOML files, ensuring consistency and accurate reflection of each model's capabilities. This includes updates for models like gpt-3.5-turbo, gpt-4, gpt-4-turbo, gpt-4o, gpt-4o-mini, gpt-5.1, gpt-5.1-codex, o1, o3, o3-mini, o3-pro, and o4-mini. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12]

2. Cost and Limit Corrections

  • Updated cost fields (input, output, cache_read) to use consistent floating-point values and corrected context/output token limits for all OpenAI models, providing more accurate pricing and usage information. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12]

3. Modalities and Input/Output Updates

  • Updated the modalities section for models to accurately reflect support for text and/or image inputs and outputs where applicable, especially for newer models with multi-modal capabilities. [1] [2] [3] [4] [5] [6] [7] [8] [9]

4. Display Name and Model Mapping Improvements

  • Overhauled model_names.json to map internal model identifiers to human-friendly display names, replacing previous 1:1 mappings. This enhances clarity for users and UIs referencing these models.

5. Model File Reorganization and Naming Consistency

  • Renamed and reorganized several model TOML files under the @cf namespace for providers like ai4bharat, aisingapore, and baai, and updated display names and metadata for BGE and Gemma models. This improves maintainability and discoverability of model files. [1] [2] [3] [4] [5] [6]

@@ -1,5 +1,4 @@
name = "GPT-OSS 20B"
family = "gpt-oss"
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

y remove family?

@rekram1-node
Copy link
Contributor

Looks like a few models had "family" dropped from them, wondering why

@superhighfives
Copy link
Contributor Author

Good catch—looks like those had been added manually. Adding back in programmatically.

@superhighfives
Copy link
Contributor Author

Great, added. 👍

@rekram1-node rekram1-node merged commit d54bc05 into sst:dev Dec 23, 2025
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants