GPT-5.6 instruction template identifies the model only as GPT-5

Description

The current models.json entries for gpt-5.6-sol, gpt-5.6-terra, and gpt-5.6-luna use an identical model_messages.instructions_template.

The template begins with wording equivalent to:

You are Codex, an agent based on GPT-5.

It does not identify the selected model as GPT-5.6.

Evidence

In the current main branch:

• All three GPT-5.6 entries have the same instruction template.
• Each template is 17,730 characters.
• Each has SHA-256:

cbefa6b0bede0e332d957fca70ccacf9f12f4c0ecdf81b819e5cbe1a3b16e265

Question

Is the generic “based on GPT-5” wording intentional for the GPT-5.6 family?

If not, could the prompt use the selected model family/version, or a model-specific variable, so that GPT-5.6 sessions do not describe themselves only as GPT-5?

This is about prompt identity wording only; the model selection itself is correctly represented by the slug, such as gpt-5.6-sol.