Problem
GPT-5.5 has been released, but hermes-agent does not recognize or route it correctly. The model is absent from the provider catalog, the Codex image-gen plugin still defaults to gpt-5.4, tests reference the old model, and documentation shows outdated model names.
Current Behavior
hermes_cli/models.py: gpt-5.5 is not listed in VERCEL_AI_GATEWAY_MODELS, not registered under _PROVIDER_MODELS for the openai, copilot, opencode-zen, or kilocode providers, and not included in _PRIORITY_PROCESSING_MODELS.
plugins/image_gen/openai-codex/__init__.py: _CODEX_CHAT_MODEL defaults to gpt-5.4, so Codex image-gen requests use a stale model.
tests/gateway/test_fast_command.py: Monkeypatched _resolve_gateway_model stubs return gpt-5.4, so gateway routing tests do not cover the current model.
tests/plugins/image_gen/test_openai_codex_provider.py: Model assertion expects gpt-5.4.
website/docs/integrations/providers.md: CLI examples and config default reference gpt-5.4.
website/docs/reference/faq.md: /model example shows gpt-5.4.
Expected Behavior
gpt-5.5 is recognized by the Vercel AI gateway and registered for all supported providers (openai, copilot, opencode-zen, kilocode).
gpt-5.5 is in the priority processing set so requests are handled with the correct routing path.
- Codex image-gen plugin uses
gpt-5.5 as its chat model.
- Tests assert against
gpt-5.5.
- Documentation reflects
gpt-5.5 as the current default.
Affected Files
hermes_cli/models.py
plugins/image_gen/openai-codex/__init__.py
tests/gateway/test_fast_command.py
tests/plugins/image_gen/test_openai_codex_provider.py
website/docs/integrations/providers.md
website/docs/reference/faq.md
Related Issues
Problem
GPT-5.5 has been released, but hermes-agent does not recognize or route it correctly. The model is absent from the provider catalog, the Codex image-gen plugin still defaults to gpt-5.4, tests reference the old model, and documentation shows outdated model names.
Current Behavior
hermes_cli/models.py:gpt-5.5is not listed inVERCEL_AI_GATEWAY_MODELS, not registered under_PROVIDER_MODELSfor the openai, copilot, opencode-zen, or kilocode providers, and not included in_PRIORITY_PROCESSING_MODELS.plugins/image_gen/openai-codex/__init__.py:_CODEX_CHAT_MODELdefaults togpt-5.4, so Codex image-gen requests use a stale model.tests/gateway/test_fast_command.py: Monkeypatched_resolve_gateway_modelstubs returngpt-5.4, so gateway routing tests do not cover the current model.tests/plugins/image_gen/test_openai_codex_provider.py: Model assertion expectsgpt-5.4.website/docs/integrations/providers.md: CLI examples and config default referencegpt-5.4.website/docs/reference/faq.md:/modelexample showsgpt-5.4.Expected Behavior
gpt-5.5is recognized by the Vercel AI gateway and registered for all supported providers (openai, copilot, opencode-zen, kilocode).gpt-5.5is in the priority processing set so requests are handled with the correct routing path.gpt-5.5as its chat model.gpt-5.5.gpt-5.5as the current default.Affected Files
hermes_cli/models.pyplugins/image_gen/openai-codex/__init__.pytests/gateway/test_fast_command.pytests/plugins/image_gen/test_openai_codex_provider.pywebsite/docs/integrations/providers.mdwebsite/docs/reference/faq.mdRelated Issues
hermes updaterunsnpm installinstead ofnpm ci, dirtying package-lock.json after every update #4048 — Codex image-gen provider