COPILOT MOVES TO GPT-5.5; CODING NOW RUNS ON OPENAI’S UNIFIED MODEL
Microsoft is rolling GPT-5.5 across Copilot while OpenAI confirms there’s no separate Codex model anymore. Microsoft says GPT-5.5 is now powering Copilot acros...
Microsoft is rolling GPT-5.5 across Copilot while OpenAI confirms there’s no separate Codex model anymore.
Microsoft says GPT-5.5 is now powering Copilot across GitHub, Microsoft 365, Copilot Studio, and Foundry with stronger reasoning and multi‑step execution, aimed at fewer prompts and more complete outputs (Moneycontrol, Newsbytes).
OpenAI’s Romain Huet says Codex was unified into the main model starting with GPT‑5.4, and GPT‑5.5 pushes further on agentic coding and computer use—so there’s no separate “-Codex” track to choose from anymore Simon Willison.
Enterprise admins shouldn’t expect instant availability everywhere: model visibility can depend on enterprise policy toggles and GHES versioning, which has already tripped up orgs enabling newer Copilot models DevActivity.
Copilot’s behavior will shift on real code and docs as GPT-5.5 handles multi-step tasks more autonomously.
No separate Codex line means simpler model selection but different tradeoffs for latency, reliability, and guardrails.
-
terminal
A/B Copilot before/after GPT-5.5 on representative services: measure suggestion acceptance, compile/test pass rates, and latency.
-
terminal
In Copilot Studio, trial multi-step agents for routine runbooks (e.g., PR triage, SQL report refresh); track human handoff rate and errors.
Legacy codebase integration strategies...
- 01.
If on GHES, verify enterprise-level Copilot policy enablement and server version alignment before expecting GPT-5.5 availability.
- 02.
Recheck DLP, repo scopes, and audit pipelines as more autonomous multi-step actions increase the blast radius of a bad suggestion.
Fresh architecture paradigms...
- 01.
Default to the unified GPT-5.5 for code and ops agents; design flows that assume deeper planning and multi-step execution.
- 02.
Bake in evals for codegen and SQL correctness early; wire CI to run fast checks on any AI-suggested changes.
Get daily MICROSOFT + SDLC updates.
- Practical tactics you can ship tomorrow
- Tooling, workflows, and architecture notes
- One short email each weekday