Skip to content

Commit f49fddd

Browse files
AceGreenmanclaude
andcommitted
preset(standard): add explicit codex default (gpt-5.4)
Previously the default tier had no codex override, so Codex users running `extends: standard` with no further overrides got the Anthropic default flattened as-is — which Codex can't route natively. Pin the default on Codex to gpt-5.4 (OpenAI's current recommended model per their 2026-04 docs) so the preset works out of the box on Codex without requiring the user to remember they need an override. Also tightens the header comment to name the specific harness/provider docs used to verify picks (Claude Code model config, Cursor models, Codex models, OpenCode models) rather than a generic list. 31/31 iso-route tests pass — preset load + merge semantics unchanged. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
1 parent 756a58d commit f49fddd

1 file changed

Lines changed: 8 additions & 3 deletions

File tree

packages/iso-route/presets/standard.yaml

Lines changed: 8 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -1,9 +1,11 @@
11
# @razroo/iso-route — standard preset
22
#
33
# Opinionated cost-tiered routing across Claude Code, Cursor, Codex, and
4-
# OpenCode. Verified against each provider's 2026-04 model catalogs:
5-
# Anthropic docs, OpenAI docs, xAI docs, Google Gemini docs, OpenCode
6-
# Zen/Go proxy listings.
4+
# OpenCode. Verified against 2026-04 harness/provider docs:
5+
# - Claude Code model config (aliases + pinned model IDs)
6+
# - Cursor models docs (manual picker; no file-based role binding)
7+
# - Codex models docs (gpt-5.4 default recommendation)
8+
# - OpenCode models docs (provider/model ID format + variants)
79
#
810
# Three roles, each with per-harness overrides:
911
# - fast — cheap + quick, for procedural work
@@ -23,6 +25,9 @@ default:
2325
provider: anthropic
2426
model: claude-sonnet-4-6
2527
targets:
28+
codex:
29+
provider: openai
30+
model: gpt-5.4
2631
opencode:
2732
provider: opencode
2833
model: opencode/glm-5.1

0 commit comments

Comments
 (0)