Meta description (≈155 chars): OpenAI’s GPT-5.2 introduces Thinking and Instant modes, expanding to paid ChatGPT, Microsoft 365 Copilot, and the OpenAI API. 

OpenAI Launches GPT-5.2: “Thinking” and “Instant” Roll Out in ChatGPT, Copilot, and the API

OpenAI has officially released GPT-5.2 on December 11, 2025, positioning it as its most advanced model family yet for professional work and long-running agent workflows. The launch introduces multiple variants—including GPT-5.2 Thinking, GPT-5.2 Instant, and a higher-end GPT-5.2 Pro—and begins rolling out across paid ChatGPT plans, Microsoft 365 Copilot, and the OpenAI API. OpenAI+2OpenAI+2

The timing is notable: the release lands amid intensified competition from Google’s Gemini 3, as the AI race shifts from flashy demos to reliability, tool-use, and “end-to-end” task completion. blog.google+1

What’s new in GPT-5.2

OpenAI describes GPT-5.2 as a major step forward in real-world knowledge work—especially for tasks like building spreadsheets, producing presentations, writing and refactoring code, handling vision inputs, and maintaining coherence across very long contexts. OpenAI+1

1) Two main modes: Instant vs. Thinking (plus Pro)

• GPT-5.2 Instant: optimized for speed and everyday tasks like explanations, how-tos, and writing/translation. OpenAI
• GPT-5.2 Thinking: tuned for deeper reasoning—coding, long-document analysis, math/logic, and structured planning. OpenAI+1
• GPT-5.2 Pro: positioned as the most “trustworthy” option for difficult questions where higher quality is worth the extra time/cost. OpenAI

2) Stronger benchmark performance (especially “knowledge work” and coding)

OpenAI highlights gains on GDPval (knowledge-work evaluation spanning 44 occupations) and SWE-Bench Pro (real-world software engineering), framing GPT-5.2 Thinking as its best “professional” model so far. OpenAI+1

3) Reliability and hallucination reduction

One of the biggest claims around GPT-5.2 is improved factual reliability. OpenAI states that for GPT-5.2 Thinking, responses containing errors were “30% (relative) less common” than GPT-5.1 Thinking on a set of de-identified ChatGPT queries (with search enabled). OpenAI

Meanwhile, multiple reports cite OpenAI briefing numbers claiming ~38% fewer hallucinations on factual-style benchmarks—so you’ll see both figures referenced depending on which evaluation set is being quoted. WIRED+1

Bottom line: GPT-5.2 is being marketed as more dependable, but OpenAI still emphasizes it’s not perfect and should be double-checked for critical decisions. OpenAI

Where GPT-5.2 is available right now

ChatGPT rollout (paid plans first)

OpenAI says GPT-5.2 (Instant, Thinking, Pro) begins rolling out starting today to paid tiers (including Plus/Pro/Business/Enterprise), with gradual deployment to keep performance stable. It also notes GPT-5.1 will remain available for a limited time under legacy models before being sunset in ChatGPT. OpenAI

Microsoft 365 Copilot integration

Microsoft confirms GPT-5.2 is coming to Microsoft 365 Copilot and Copilot Studio, explicitly calling out Thinking for complex problems and Instant for everyday writing and translation. Microsoft

API availability, naming, and pricing

For developers, OpenAI lists model IDs such as:

• gpt-5.2 (Thinking)
• gpt-5.2-chat-latest (Instant)
• gpt-5.2-pro (Pro)

OpenAI also publishes updated token pricing for GPT-5.2 and Pro in its release post. OpenAI+1

Safety notes: more focus on sensitive conversations

OpenAI says GPT-5.2 builds on its “safe completion” approach and includes targeted improvements for handling sensitive prompts involving self-harm, mental health distress, and emotional reliance. It also mentions early rollout of an age prediction model to apply protections for users under 18. OpenAI

Why this launch matters: OpenAI vs. Gemini 3

Google’s Gemini 3 announcement and related product pushes have increased pressure across the frontier-model market. OpenAI’s GPT-5.2 launch is widely framed as a direct response—especially with the “code red” narrative circulating in reporting about internal urgency to accelerate ChatGPT improvements. Reuters+2blog.google+2

From a user perspective, the competition is increasingly about:

• tool reliability (agents that actually finish workflows),
• long-context accuracy (multi-document reasoning),
• and lower error rates (fewer confident wrong answers).

GPT-5.2 is OpenAI’s attempt to claim leadership across those dimensions. OpenAI+2OpenAI+2

What to watch next

A few practical questions will determine whether GPT-5.2 truly “sticks”:

• Does it feel consistently more reliable in day-to-day use (not just on benchmarks)?
• How quickly does rollout reach all paid users globally?
• How do developers adapt to new pricing vs. quality gains?
• How fast do Google and others respond with updates and aggressive bundling?

For now, GPT-5.2 is clearly being positioned as OpenAI’s most serious “work model” yet—built not just to chat, but to execute multi-step tasks with fewer mistakes.