# NousResearch/hermes-agent · Codex provider: gpt-5.3-codex-spark model availability and selection ## Open now - Nothing open on GitHub. The members concern making the gpt-5.3-codex-spark model (and, more broadly, Codex CLI-only models) available or selectable in the Codex/openai-codex provider, including for ChatGPT Pro. They include a bug report about being unable to select the model and several pull requests that add or restore it; their outcomes are not stated in the titles. ## TL;DR - 4 member(s): 0 open issue(s), 1 closed issue(s), 0 open PR(s), 1 merged PR(s), 2 PR(s) closed without merging. - A fix was merged to main in PR #22991. Verify it addresses your case. - **Recommended:** read PR #22991 (and any competing PRs below) before filing. Confirm the merged change covers your symptom. ## Maintainer resolution verdict - Internal-tracker verdict over this cluster's issue members: `resolved` · deciding PR(s): `pr-22991`. This is distinct from the fix status above. Weigh it yourself. ## How this cluster was grouped - Link evidence (`STRUCTURAL_SPANNING`): every member is connected by structural evidence: shared changed files, or a closing/superseded reference. - Pairwise link density `0.50`: the share of member pairs that carry a link, counting a structural link in full and a relevance-only link at 0.36. `1.00` means every pair is linked. ## Closed / duplicate issues - #16172: \[Bug\]: Cannot select gpt-5.3-codex-spark in openai-codex provider ## Prior & competing PRs - **#22991: fix(codex): restore gpt-5.3-codex-spark for ChatGPT Pro (salvage #18286 + #19530, fixes #16172)**. merged to main · fix tier `verified` - maintainer direction: category `merged` - files: `agent/model_metadata.py`, `hermes_cli/codex_models.py`, `hermes_cli/model_switch.py`, `scripts/release.py`, `tests/agent/test_model_metadata.py`, `tests/hermes_cli/test_codex_cli_model_picker.py`, `tests/hermes_cli/test_codex_models.py`, `tests/hermes_cli/test_openai_codex_model_validation_fallback.py` - fix hunks: - `agent/model_metadata.py` ```diff @@ -157,6 +157,13 @@ def _strip_provider_prefix(model: str) -> str: "gpt-5.4-nano": 400000, # 400k (not 1.05M like full 5.4) "gpt-5.4-mini": 400000, # 400k (not 1.05M like full 5.4) "gpt-5.4": 1050000, # GPT-5.4, GPT-5.4 Pro (1.05M context) + # gpt-5.3-codex-spark is Codex-OAuth-only (ChatGPT Pro entitlement) and + # uses a smaller 128k window than other gpt-5.x slugs. Listed here as + # a defensive override so the longest-substring fallback doesn't match + # the generic "gpt-5" entry below (400k) and report the wrong limit if + # Spark's context ever needs to be resolved through this path. Real + # usage flows through _CODEX_OAUTH_CONTEXT_FALLBACK at line ~1113. + "gpt-5.3-codex-spark": 128000, "gpt-5.1-chat": 128000, # Chat variant has 128k context "gpt-5": 400000, # GPT-5.x base, mini, codex variants (400k) "gpt-4.1": 1047576, ``` ```diff @@ -1108,6 +1115,12 @@ def _query_anthropic_context_length(model: str, base_url: str, api_key: str) -> "gpt-5.1-codex-max": 272_000, "gpt-5.1-codex-mini": 272_000, "gpt-5.3-codex": 272_000, + # Spark runs on specialised low-latency hardware and exposes a smaller + # 128k window than other Codex OAuth slugs. Listed explicitly so the + # longest-key-first fallback resolves it correctly — substring match + # on "gpt-5.3-codex" otherwise wins and reports 272k. Availability is + # gated by ChatGPT Pro entitlement on the Codex backend. + "gpt-5.3-codex-spark": 128_000, "gpt-5.2-codex": 272_000, "gpt-5.4-mini": 272_000, "gpt-5.5": 272_000, ``` - `hermes_cli/codex_models.py` ```diff @@ -16,6 +16,19 @@ "gpt-5.4-mini", "gpt-5.4", "gpt-5.3-codex", + # gpt-5.3-codex-spark is in research preview and is exposed *only* via + # the Codex CLI / OAuth backend (chatgpt.com/backend-api/codex/models) + # for ChatGPT Pro subscribers. It is NOT available in the public OpenAI + # API, so it intentionally stays out of the "openai" provider catalog + # in hermes_cli/models.py — only the openai-codex (OAuth) provider + # surfaces it. The Codex backend reports ``supported_in_api: false`` for + # this slug; that flag describes API availability, not Codex backend + # availability, so the fetch/cache code paths below intentionally do + # not filter on it. PR #12994 removed this entry on the assumption it + # was unsupported — that was wrong; restored here. Keep it in the + # curated fallback so Pro users still see Spark in `/model` when live + # discovery is unavailable (offline first run, transient API failure). + "gpt-5.3-codex-spark", "gpt-5.2-codex", "gpt-5.1-codex-max", "gpt-5.1-codex-mini", ``` ```diff @@ -26,6 +39,11 @@ ("gpt-5.4-mini", ("gpt-5.3-codex", "gpt-5.2-codex")), ("gpt-5.4", ("gpt-5.3-codex", "gpt-5.2-codex")), ("gpt-5.3-codex", ("gpt-5.2-codex",)), + # Surface Spark whenever any compatible Codex template is present so + # accounts hitting the live endpoint with an older lineup still see + # Spark in the picker. Backend gates real availability by ChatGPT Pro + # entitlement; Hermes does not. + ("gpt-5.3-codex-spark", ("gpt-5.3-codex", "gpt-5.2-codex")), ] ``` ```diff @@ -78,8 +96,10 @@ def _fetch_models_from_api(access_token: str) -> List[str]: if not isinstance(slug, str) or not slug.strip(): continue slug = slug.strip() - if item.get("supported_in_api") is False: - continue + # Codex CLI's catalog uses ``supported_in_api`` for the public OpenAI + # API, not for the OAuth-backed Codex backend that this provider uses. + # Some valid Codex CLI models (for example gpt-5.3-codex-spark) are + # marked false here but are still accepted by the Codex route. visibility = item.get("visibility", "") if isinstance(visibility, str) and visibility.strip().lower() in ("hide", "hidden"): continue ``` ```diff @@ -128,8 +148,9 @@ def _read_cache_models(codex_home: Path) -> List[str]: if not isinstance(slug, str) or not slug.strip(): continue slug = slug.strip() - if item.get("supported_in_api") is False: - continue + # Do not filter on ``supported_in_api`` here. It describes the + # public OpenAI API, while Hermes openai-codex talks to the same + # OAuth-backed Codex backend as Codex CLI. visibility = item.get("visibility") if isinstance(visibility, str) and visibility.strip().lower() in ("hide", "hidden"): continue ``` - `hermes_cli/model_switch.py` ```diff @@ -1342,7 +1342,14 @@ def _has_aws_sdk_creds_for_listing(slug: str) -> bool: if not has_creds: continue - if hermes_slug in {"copilot", "copilot-acp"}: + if hermes_slug in {"openai-codex", "copilot", "copilot-acp"}: + # Use live OAuth-backed discovery so the gateway /model picker + # matches what the user's authenticated Codex/Copilot backend + # actually serves — including ChatGPT-Pro-only Codex slugs + # (e.g. gpt-5.3-codex-spark) that aren't in the static curated + # catalog. ``provider_model_ids()`` falls back to the curated + # list when the live endpoint is unreachable, so this is safe + # for unauthenticated and offline cases too. model_ids = provider_model_ids(hermes_slug) # For aws_sdk providers (bedrock), use live discovery so the list # reflects the active region (eu.*, ap.*) not the static us.* list. ``` - `scripts/release.py` ```diff @@ -235,6 +235,9 @@ "[redacted:email]": "Ito-69", "[redacted:email]": "georgeglessner", "[redacted:email]": "maxims-oss", + # Codex Spark restoration salvage (May 2026) + "[redacted:email]": "nederev", + "[redacted:email]": "askclaw-vesper", "[redacted:email]": "nazirulhafiy", "CREWorx@users.noreply.github.com": "BadTechBandit", "[redacted:email]": "Yoimex", ``` - `tests/agent/test_model_metadata.py` ```diff @@ -262,8 +262,9 @@ def test_dict_is_not_empty(self): class TestCodexOAuthContextLength: """ChatGPT Codex OAuth imposes lower context limits than the direct OpenAI API for the same slugs. Verified Apr 2026 via live probe of - chatgpt.com/backend-api/codex/models: every model returns 272k, while + chatgpt.com/backend-api/codex/models: most models return 272k, while models.dev reports 1.05M for gpt-5.5/gpt-5.4 and 400k for the rest. + (Known exception: gpt-5.3-codex-spark is 128k.) """ def setup_method(self): ``` ```diff @@ -277,25 +278,28 @@ def test_fallback_table_used_without_token(self): """ from agent.model_metadata import get_model_context_length + expected = { + "gpt-5.5": 272_000, + "gpt-5.4": 272_000, + "gpt-5.4-mini": 272_000, + "gpt-5.3-codex": 272_000, + "gpt-5.3-codex-spark": 128_000, + "gpt-5.2-codex": 272_000, + "gpt-5.1-codex-max": 272_000, + "gpt-5.1-codex-mini": 272_000, + } + with patch("agent.model_metadata.get_cached_context_length", return_value=None), \ patch("agent.model_metadata.save_context_length"): - for model in ( - "gpt-5.5", - "gpt-5.4", - "gpt-5.4-mini", - "gpt-5.3-codex", - "gpt-5.2-codex", - "gpt-5.1-codex-max", - "gpt-5.1-codex-mini", - ): + for model, expected_ctx in expected.items(): ctx = get_model_context_length( model=model, base_url="https://chatgpt.com/backend-api/codex", api_key="", provider="openai-codex", ) - assert ctx == 272_000, ( - f"Codex {model}: expected 272000 fallback, got {ctx} " + assert ctx == expected_ctx, ( + f"Codex {model}: expected {expected_ctx} fallback, got {ctx} " "(models.dev leakage?)" ) ``` - `tests/hermes_cli/test_codex_cli_model_picker.py` ```diff @@ -75,6 +75,37 @@ def test_normal_path_still_works(hermes_auth_only_env): assert "openai-codex" in slugs +def test_codex_picker_uses_live_codex_catalog(hermes_auth_only_env, tmp_path, monkeypatch): + """The gateway /model picker should surface Codex CLI-only listed models.""" + from hermes_cli.model_switch import list_authenticated_providers + + codex_home = tmp_path / "codex-home" + codex_home.mkdir() + (codex_home / "models_cache.json").write_text(json.dumps({ + "models": [ + {"slug": "gpt-5.5", "priority": 0, "supported_in_api": True}, + {"slug": "gpt-5.3-codex-spark", "priority": 7, "supported_in_api": False}, + ] + })) + monkeypatch.setenv("CODEX_HOME", str(codex_home)) + # Force the cache fallback path — without this the test issues a real + # 10s HTTP probe to chatgpt.com/backend-api/codex/models which is both + # slow and non-deterministic in CI/sandboxed environments. + monkeypatch.setattr( + "hermes_cli.codex_models._fetch_models_from_api", + lambda access_token: [], + ) + + providers = list_authenticated_providers( + current_provider="openai-codex", + max_models=10, + ) + + codex = next(p for p in providers if p["slug"] == "openai-codex") + assert "gpt-5.3-codex-spark" in codex["models"] + assert codex["total_models"] == len(codex["models"]) + + @pytest.fixture() def claude_code_only_env(tmp_path, monkeypatch): """Set up an environment where Anthropic credentials only exist in ``` - `tests/hermes_cli/test_codex_models.py` ```diff @@ -1,10 +1,6 @@ import json -import os -import sys from unittest.mock import patch -sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..")) - from hermes_cli.codex_models import DEFAULT_CODEX_MODELS, get_codex_model_ids ``` ```diff @@ -17,6 +13,7 @@ def test_get_codex_model_ids_prioritizes_default_and_cache(tmp_path, monkeypatch { "models": [ {"slug": "gpt-5.3-codex", "priority": 20, "supported_in_api": True}, + {"slug": "gpt-5.3-codex-spark", "priority": 6, "supported_in_api": False}, {"slug": "gpt-5.1-codex", "priority": 5, "supported_in_api": True}, {"slug": "gpt-5.4", "priority": 1, "supported_in_api": True}, {"slug": "gpt-5-hidden-codex", "priority": 2, "visibility": "hidden"}, ``` ```diff @@ -31,6 +28,9 @@ def test_get_codex_model_ids_prioritizes_default_and_cache(tmp_path, monkeypatch assert models[0] == "gpt-5.2-codex" assert "gpt-5.1-codex" in models assert "gpt-5.3-codex" in models + # Codex CLI marks Spark unsupported in the public API, but the Codex + # backend still accepts it via the OAuth-backed CLI/Hermes route. + assert "gpt-5.3-codex-spark" in models # Non-codex-suffixed models are included when the cache says they're available assert "gpt-5.4" in models assert "gpt-5.4-mini" in models ``` ```diff @@ -54,7 +54,7 @@ def test_get_codex_model_ids_falls_back_to_curated_defaults(tmp_path, monkeypatc assert models[: len(DEFAULT_CODEX_MODELS)] == DEFAULT_CODEX_MODELS assert "gpt-5.4" in models - assert "gpt-5.3-codex-spark" not in models + assert "gpt-5.3-codex-spark" in models def test_get_codex_model_ids_adds_forward_compat_models_from_templates(monkeypatch): ``` ```diff @@ -65,7 +65,49 @@ def test_get_codex_model_ids_adds_forward_compat_models_from_templates(monkeypat models = get_codex_model_ids(access_token="[redacted:secret-assignment]") - assert models == ["gpt-5.2-codex", "gpt-5.4-mini", "gpt-5.4", "gpt-5.3-codex"] + assert models == [ + "gpt-5.2-codex", + "gpt-5.4-mini", + "gpt-5.4", + "gpt-5.3-codex", + "gpt-5.3-codex-spark", + ] + + +def test_fetch_from_api_keeps_supported_in_api_false_models(monkeypatch): + """Regression: gpt-5.3-codex-spark is returned by the live Codex backend + with ``supported_in_api: false`` because it isn't in the public OpenAI + API. The Codex CLI / OAuth route still serves it for ChatGPT Pro + accounts, so we must not drop it on that flag. visibility=hidden is + the separate signal that *should* still filter entries out. + """ + import sys + from hermes_cli import codex_models + + class _FakeResp: + status_code = 200 + + def json(self): + return { + "models": [ + {"slug": "gpt-5.5", "priority": 0, "supported_in_api": True}, + {"slug": "gpt-5.3-codex-spark", "priority": 7, "supported_in_api": False}, + {"slug": "gpt-5-internal", "priority": 99, "visibility": "hidden"}, + ] + } + + class _FakeHttpx: + @staticmethod + def get(url, headers=None, timeout=None): + return _FakeResp() + + monkeypatch.setitem(sys.modules, "httpx", _FakeHttpx) + + models = codex_models._fetch_models_from_api(access_token="tok") + + assert "gpt-5.5" in models + assert "gpt-5.3-codex-spark" in models + assert "gpt-5-internal" not in models def test_model_command_uses_runtime_access_token_for_codex_list(monkeypatch): ``` - `tests/hermes_cli/test_openai_codex_model_validation_fallback.py` ```diff @@ -1,9 +1,18 @@ """Regression tests for OpenAI Codex model validation when the listing lags behind actually usable backend model IDs. -The bug: `/model` and `switch_model()` reject `gpt-5.3-codex-spark` because the -OpenAI Codex listing omits it, even though direct runtime calls with -`--provider openai-codex -m gpt-5.3-codex-spark` succeed. +The bug originally reported in #16172: `/model` and `switch_model()` rejected +`gpt-5.3-codex-spark` because the curated listing omitted it, even though direct +runtime calls succeeded. PR #19729 fixed this by soft-accepting unknown-but- +plausible Codex slugs with a warning, and this test pins the soft-accept +behavior so it doesn't regress. + +Note: gpt-5.3-codex-spark itself is now in the curated catalog (PR #22991), +so the real-world Spark request takes the `recognized=True` fast path. This +test still uses Spark as the example slug but explicitly mocks +``provider_model_ids`` to omit it, exercising the soft-accept path generically +for any future entitlement-gated Codex slug that ships before Hermes catalogs +it. """ from unittest.mock import patch ``` - **#18286: feat(codex): add gpt-5.3-codex-spark model**. closed without merge - maintainer direction: category `merged_via`; implication `superseded`; refs `44cdf555a`, `dcc8de83a`, `2a6f3deb5`, `c6dc295a3`, `#22991`, `#19530`, `#12994`, `#16172` - **#19530: fix: surface Codex CLI-only models**. closed without merge - maintainer direction: category `merged_via`; implication `superseded`; refs `44cdf555a`, `#22991`, `#18286`, `#12994`, `#16172` ## Fix status - Merged to main in PR #22991 (cited above). No release data is tracked. Verify downstream. - Fix last verified (merged) `2026-05-10`.