Repository navigation
ug: Use GPT reasoning levels for GPT models Codex doesn't bundle - #1034
Draft
max-rozen-oss-db wants to merge 3 commits into
Draft
max-rozen-oss-db wants to merge 3 commits into
max-rozen-oss-db wants to merge 3 commits into
Conversation
The fallback catalog entry for GPT models missing from Codex's bundle used the non-GPT hosted defaults, which only offer reasoning effort none. The gateway rejects none for gpt-5-nano and gpt-5-mini, so every request failed. GPT fallbacks now offer low/medium/high with medium as the default, and the warning no longer promises that ug codex update adds the metadata. Co-authored-by: Isaac <no-reply@databricks.com>
Co-authored-by: Isaac <no-reply@databricks.com>
With no default_reasoning_level Codex omits reasoning.effort, so the gateway applies each model's default (medium for gpt-5-nano/mini, none for gpt-5.4-nano) instead of a value UG guessed. low/medium/high stay selectable. Co-authored-by: Isaac <no-reply@databricks.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Fixes the #ai-gateway report where
ug codexwith managedsystem.ai.gpt-5-nano/gpt-5-4-nano/gpt-5-minifailed every turn withUnsupported value: 'none' is not supported with the 'gpt-5-nano' model.Current Codex releases don't bundle these older GPT models, so
build_codex_catalogbuilt their entries from the non-GPT hosted defaults, which only offer reasoning effortnone. Codex then sendsreasoning.effort: "none", which the gateway rejects for GPT-5 base models.ug codex updatecan't help because no Codex release adds them.default_reasoning_level, so Codex omitsreasoning.effortand the gateway applies the model's own default (medium for gpt-5-nano/mini, none for gpt-5.4-nano). low/medium/high stay selectable; every GPT-5-generation model accepts them.ug codex updatefixes it.Verification
noneand the turn fails with the reported 400.databricks-gpt-5-nanohas no default and low/medium/high;codex execwith it completes a turn. Direct gateway calls withoutreasoningreturn 200 for all three models.test_codex_catalog.py,test_agent_codex.py,test_codex_smart_routing_v2.py, and the integration contract tests pass locally./modelpicker. Codex converts a null default tononewhen building picker presets, so that path should be checked by hand.This pull request and its description were written by the Isaac LLM tool (Claude Code, session
053e9e2b-b4dd-41dc-9287-b081cc6ed1bc).