Model guidance is stale: GPT-5.6 family (Sol/Terra/Luna) shipped, superseding GPT-5.5/5.4

[BaseInfinity]

Verified: model guidance is stale — GPT-5.6 family shipped

Fact-checked directly against OpenAI's own docs (not inferred): the flagship model has moved from GPT-5.5 to a new GPT-5.6 family with three tiers.

Model	Positioning	Price (in/out per 1M)
gpt-5.6-sol	Flagship — "strongest capability for complex coding, computer use, research, and cybersecurity"	$5 / $30
gpt-5.6-terra	Balanced — "competitive with GPT-5.5 at a lower cost"	$2.50 / $15
gpt-5.6-luna	High-volume/cheap — "fast and affordable... lowest cost in the family"	$1 / $6

Sources: https://developers.openai.com/api/docs/guides/latest-model.md, https://learn.chatgpt.com/docs/models, https://developers.openai.com/api/docs/pricing. Reasoning-effort levels (none/low/medium/high/xhigh/max) are unchanged; OpenAI's own guidance is to preserve the prior GPT-5.5 effort setting as baseline, test one level lower, and reserve max for the hardest quality-first work.

Why this repo specifically

Same shape as codex-sdlc-wizard#60, verified directly against this repo's own content (not assumed from the sibling):

• AGENTS.md (lines 6-7): repo-wide reasoning policy hardcodes gpt-5.5 xhigh as the mandatory profile for all wizard-repo work.
• SKILL.md (line 24): same policy repeated — "always stays on maximum (gpt-5.5 xhigh throughout)."
• README.md (~10 refs): codex -m gpt-5.5 ... invocation examples, review_model = "gpt-5.5" config example, the mixed/maximum profile definitions (gpt-5.4-mini + gpt-5.5 xhigh review), and the Setup A/B/C lane table — this repo doesn't have a standalone AI_SETUP_LANES.md, the lane guidance lives directly in README.md.

All three tiers are candidates for remapping onto the new family (Sol for the flagship/review role, Terra for a saver-tier driver, Luna for the cheap/mini tier) — but same caveat as codex-sdlc-wizard#60: whether Terra/Luna genuinely replace the old GPT-5.3 Spark / GPT-5.4-mini tiers functionally (not just by rough price analogy) needs verification, not just a name swap.

Proposal

Same recipe as codex-sdlc-wizard#60 and already proven on claude-sdlc-wizard (AI_SETUP_LANES.md's v3 update, PR #441, Fable + 3-round Codex xhigh CERTIFIED): a scoped Fable pass on AGENTS.md, SKILL.md, and README.md to remap onto the GPT-5.6 family, verified against current docs/pricing, then a Codex cross-model review of the change itself before merging model-selection guidance that every consumer of this wizard inherits.

---

Submitted as part of a cross-repo self-enforcement/currency review pass (see claude-sdlc-wizard#439 -> PR #441, codex-sdlc-wizard#60).