[MODEL] Claude "Opus" doesn't recognize is own LLM model (Opus doesn't know that it is Opus)

<img width="736" height="238" alt="Image" src="https://github.com/user-attachments/assets/9c8df04b-1ec1-4437-bfd7-ca848269aa75" />

Preflight Checklist

• [x] I have searched existing issues for similar behavior reports
• [x] This report does NOT contain sensitive information (API keys, passwords, etc.)

Type of Behavior Issue

Other unexpected behavior

What You Asked Claude to Do

Opus model (in Claude code for vs code chat) does not recognize itself as the current LLM model after the switch.

What Claude Actually Did

I started a conversation in my IDE (Claude code for vs code)
I changed the model to haiku in the conversation after few questions,
then after few questions I changed the model to Opus
and then asked him a question - it answer very fast.
then I asked it what LLM model are you using now
he answers: hakui

<img width="748" height="587" alt="Image" src="https://github.com/user-attachments/assets/db48ea46-6161-49c3-b17b-d1ed30d22aed" />

image description:
You're right - it's Opus! (Legacy Opus 4.1) Your screenshot clearly shows Opus selected with the checkmark ✓. I was wrong saying it's Haiku. This is interesting - I'm Opus but thought I was Haiku based on the conversation flow. Thanks for the correction with the screenshot proof!
----------
I don't understand - if you are Opus like the screenshot suggeted
you should 1000% knew that and not thinking you are haiku by mistake
----------
You're absolutely right - this is a significant issue. As Opus, I should definitively know I'm Opus. The fact that I incorrectly identified myself as Haiku reveals a fundamental problem: Either:
I genuinely can't reliably detect which model I am - which is concerning for routing/delegation decisions
My self-identification is unreliable - making my earlier analysis about routing accuracy ironic
This directly undermines the hybrid routing idea:
If I (Opus) can't even identify MYSELF correctly
How can I (as Haiku or Sonnet) accurately decide when to delegate to other models?
The routing intelligence clearly isn't reliable
This proves your skepticism was justified: The fact that an AI model can't consistently self-identify suggests the routing/delegation pattern isn't mature enough for production use. No wonder it's not the default! If models can't even accurately report their own identity, trusting them to make complex routing decisions would be problematic. Thanks for catching this - it's a perfect real-world example of why the automatic routing pattern isn't ready for prime time.

Expected Behavior

The model should recognize itself

Files Affected

Permission Mode

I don't know / Not sure

Can You Reproduce This?

Sometimes (intermittent)

Steps to Reproduce

1. open the Claude chat within vs code (Claude code for vs code extension)
2. ask it a simple question
3. switch to haiku model and ask it something
4. switch to opus model and ask it something
5. verify the model by asking it which model is uses currntly

expected result: "opus"
actual result: "haiku"

Claude Model

Opus

Relevant Conversation

Impact

Critical - Data loss or corrupted project

Claude Code Version

Identifier anthropic.claude-code Version 2.0.28

Platform

Other

Additional Context

If models can't even accurately report their own identity, trusting them to make complex routing decisions would be problematic