[BUG] Unknown model: gpt-4.1 when on MacBook Pro 16inch

[SylarRuby]

Basic checks

• I searched existing issues - this hasn't been reported
• I can reproduce this consistently
• This is a RubyLLM bug, not my application code

What's broken?

I have two macbooks, one late 2022 and an M4 Pro Chip, 16 icnch. On old macbook I have no issues but on this new mac, Im getting Unknown model: gpt-4.1:

chat = RubyLLM.chat(model: "gpt-4.1").with_tool(Story::ModifyTool.new(story: @story))
[..]

Yes, my api key is the same. On the new mac, with the issue, if I have the api key blank I get same error. Im out of ideas being as it's the same repo on both macs.

How to reproduce

# initializer
RubyLLM.configure do |config|
  config.openai_api_key = ENV["OPENAI_API_KEY"]

  # Use the new association-based acts_as API (recommended)
  config.use_new_acts_as = true
end

# run with
chat = RubyLLM.chat(model: "gpt-4.1").with_tool(Story::ModifyTool.new(story: @story))
[..]

I also tried in terminal:

chat = RubyLLM.chat

The it errors with '<main>': Unknown model: gpt-5-nano (RubyLLM::ModelNotFoundError)

Expected behavior

Should return a chat response

What actually happened

It errors with: Unknown model: gpt-4.1

Environment

• Ruby: 4.0.1
• Rails: 8.1.2
• RubyLLM: 1.11.0
• OS: MacBook 16 inch, M4 Pro Chip

[SylarRuby]

More finding. Should I call RubyLLM.models.refresh! first, it works but I only have to do this on the new MacBook Pro M4

[SylarRuby]

I've found a workaround:

RubyLLM.configure do |config|
  config.openai_api_key = ENV["OPENAI_API_KEY"]

  config.use_new_acts_as = true
end

# Had to add this
RubyLLM.models.refresh!

But I'll leave this ticket open as I feel there's a bug.

[crmne]

Closing as resolved/outdated.

Current model registry includes gpt-4.1. For environments that still have stale model data, refreshing the model registry is the right fix path.