fix(codex): remove dead gpt-5.3-codex from curated fallback list (#52492)

[liuhao1024]

What does this PR do?

Removes the dead gpt-5.3-codex model from the curated Codex fallback list and forward-compat templates. The chatgpt.com Codex backend now returns HTTP 400 for this model on ChatGPT Pro accounts, matching the same pattern that previously killed gpt-5.2-codex, gpt-5.1-codex-max, and gpt-5.1-codex-mini. Without this fix, users who are offline or hit a transient API failure see gpt-5.3-codex in /model, select it, and get a 400 crash.

Related Issue

Fixes #52492

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)

Changes Made

• hermes_cli/codex_models.py: Remove gpt-5.3-codex from DEFAULT_CODEX_MODELS; update the removal comment to include it alongside the previously removed models; update _FORWARD_COMPAT_TEMPLATE_MODELS to remove gpt-5.3-codex as a forward-compat trigger and re-anchor gpt-5.3-codex-spark on gpt-5.4/gpt-5.5 templates instead.
• tests/hermes_cli/test_codex_models.py: Update test_get_codex_model_ids_adds_forward_compat_models_from_templates to use gpt-5.5 as the trigger model (since gpt-5.3-codex is no longer a template anchor) and adjust expected output.

How to Test

1. Run python -m pytest tests/hermes_cli/test_codex_models.py -q — all 19 tests should pass.
2. Run python -m pytest tests/test_empty_model_fallback.py -q — all 14 tests should pass.
3. Verify gpt-5.3-codex is no longer in DEFAULT_CODEX_MODELS by inspecting hermes_cli/codex_models.py.

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform: macOS

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — or N/A
• I've updated cli-config.yaml.example if I added/changed config keys — or N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — or N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide — or N/A
• I've updated tool descriptions/schemas if I changed tool behavior — or N/A

Code Intelligence

• Analyzed: hermes_cli/codex_models.py (callers: models.py, model_setup_flows.py, test_codex_models.py)
• Blast radius: LOW — data-only change in curated fallback list; live discovery path unaffected
• Related patterns: Same removal pattern as gpt-5.2-codex / gpt-5.1-codex-max / gpt-5.1-codex-mini (2026-05-27)

[teknium1]

Thanks for tracing the Codex 400 report and removing the dead slug from the curated catalog.

Problems

• cli.py:5338 still directly falls back to gpt-5.3-codex when get_codex_model_ids() raises; tests/hermes_cli/test_codex_models.py:367-394 explicitly preserves that behavior. This leaves an automatic unsupported-model path after the catalog change.
• The updated test fixture at tests/hermes_cli/test_codex_models.py:63 uses gpt-5.5, but current hermes_cli/codex_models.py:55-60 synthesizes all six GPT-5.6 variants from that input. The PR's two-item expected result is therefore stale against current main.

Suggested changes

• Replace the direct exception fallback in cli.py:5338 with a supported Codex fallback and update its regression test.
• Preserve the current GPT-5.6 forward-compat behavior while asserting that gpt-5.3-codex is no longer emitted and Spark is still synthesized from supported templates.

Automated hermes-sweeper review.

[teknium1]

On current main, gpt-5.5 triggers the six GPT-5.6 forward-compat templates in hermes_cli/codex_models.py:55-60, so the two-item expectation below is stale. Update this test to preserve the GPT-5.6 synthesis behavior while asserting the removed gpt-5.3-codex slug is absent.