Creating a JSON Schema for Life-Cycle of the Models

[Anselmoo]

Code of Conduct

• I have read and agree to the GitHub Docs project's Code of Conduct

What article on docs.github.com is affected?

https://docs.github.com/en/copilot/reference/ai-models/supported-models and the family of https://docs.github.com/en/copilot/reference/ai-models/ are very valuable resources. Unfortunately, I have to parse them to keep my MCP for Guidelines up to date, as proposed in Anselmoo/mcp-ai-agent-guidelines#441. If there is a way to avoid this and read them directly as JSON schema, that would be great.

What part(s) of the article would you like to see updated?

A permalink for a GitHub api to see which models are currently supported or end of lifetime.

Additional information

No response

[welcome]

Thanks for opening this issue. A GitHub docs team member should be by to give feedback soon. In the meantime, please check out the contributing guidelines.

[rambounique0-bot]

Jesus Christ is peace your saved I'm balanced light and dark both by himself y'all are saved

[Sharra-writes]

@Anselmoo I'm trying to work out whether this is actually a documentation question or not. How would we need to update the docs? Is there a JSON schema you want to link to or is that something that would need to be created? If the latter, that's not something our team does, unfortunately. If the former, I can look into how often we link to JSON schema and under what circumstances, which is a documentation question.

[Anselmoo]

@Sharra-writes Thank you for your response and sorry that my initial post was not totally clear.

Give me the chance to clarify my request more broadly—this isn't just about my specific project, but perhaps about a growing need in the AI tooling ecosystem overall.

The Core Problem

The GitHub Copilot documentation pages (supported-models, model-comparison, model-hosting) contain valuable structured data that many tools need to consume:

Data Available	Currently Accessible As
Model names and identifiers	HTML tables
Task-based recommendations	Prose + tables
Mode support (agent, reasoning, vision)	HTML tables
Premium request multipliers	HTML tables
Client availability (VS Code, JetBrains, etc.)	HTML tables
Plan availability (Free, Pro, Business, Enterprise)	HTML tables
Hosting providers and data commitments	Prose
Retirement dates and alternatives	HTML tables

The issue at hand is that this data is exclusively accessible in a human-readable and visually appealing HTML format. From a user perspective, this is highly advantageous, as results are quickly findable and the desired locations are easily identifiable. Conversely, for my projects and potentially other initiatives that necessitate tools for web scraping and parsing, this approach is fragile, prone to errors, and does not scale effectively when models undergo frequent modifications.

Who might benefit from Machine-Readable Model Data?

As previously stated, this matter extends beyond merely my project; it is anticipated that this initiative will generally enhance the quality of other LLM initiatives, thereby benefiting several categories of tools.

The Request and Two Possible Approaches

Option A: Machine-Readable Data File (Preferred)

A JSON or YAML file hosted alongside the documentation, similar to how other GitHub resources are provided:

# Example: copilot-models.yaml
version: "2025-12-20"
models:
  - id: "gpt-5"
    name: "GPT-5"
    provider: "openai"
    status: "ga"  # ga | preview | deprecated
    modes:
      chat: true
      agent: true
      completions: false
      reasoning: true
      vision: false
    multiplier: 1.0
    plans: [free, pro, business, enterprise]
    clients: [vscode, jetbrains, github.com, cli]
    taskAreas: ["deep-reasoning", "debugging"]
    docs: "https://cdn.openai.com/gpt-5-system-card.pdf"

This could be:

• A static file in the github/docs repository (e.g., data/copilot-models.json)
• A static code section in the docs itself for copy and paste
• A permalink URL that tools can fetch (e.g., https://docs.github.com/data/copilot-models.json)

Option B: Documentation Links to Existing Resources

If creating/maintaining a data file is out of scope for the docs team, perhaps the documentation could link to:

• An internal GitHub API endpoint (if one exists)
• A canonical source of model data (if maintained elsewhere)

Precedent: Models.dev

There is already community acknowledgment of this requirement. The models.dev project offers an open API for AI model specifications (https://models.dev/api.json). However, it pertains to general AI models and does not specifically encompass GitHub Copilot's curated list, which includes distinctive metadata (such as plans, clients, multipliers, and modes).

Summary

This may matter because, as shown in the "Model retirement history" table, AI models are frequently added and deprecated. Having a machine-readable format ensures that tools remain accurate and up to date automatically, avoiding broken scrapers when HTML structures change. It also fosters ecosystem growth by lowering barriers to building Copilot-aware tools and by improving the developer experience by enabling better model selection in IDE extensions and agents.

I hope that this answer is helpful to you.

[Sharra-writes]

@Anselmoo Thanks for clarifying! I don't know if this would be a docs team initiative or something directed towards our engineers (who, unfortunately, probably don't have room for additional priorities at the moment). I'll ask and see if we have any resources for something like this.

[Anselmoo]

@Anselmoo Thanks for clarifying! I don't know if this would be a docs team initiative or something directed towards our engineers (who, unfortunately, probably don't have room for additional priorities at the moment). I'll ask and see if we have any resources for something like this.

And if the engineers have questions, they now know how to contact me.

[github-actions]

A stale label has been added to this issue, because it has been open for 30 days with no activity. If you think this issue should remain open, please add a new comment.

[subatoi]

@Anselmoo Please accept my apologies for the late conclusion here

Since this amounts to a feature request, I will pass this to the relevant team internally; however, I'm unable to guarantee a specific timeframe for a resolution or action.

I will note that we do publish https://github.com/github/docs/blob/main/data/tables/copilot/model-release-status.yml, which is somewhat related to your request.

Thank you very much for your interest in the GitHub docs