GPT-5.2
GPT-5.2 by OpenAI is the professional knowledge work frontier model with three variants: Instant, Thinking, and Pro, plus a Codex-optimized coding release.
Model Overview
Capabilities, design details, and architectural traits
GPT-5.2 - Frontier Model for Professional Knowledge Work
GPT-5.2 is OpenAI's professional knowledge work flagship, explicitly designed around the economic output of knowledge workers.
Three-Variant Structure and the reasoning.effort='none' Default
GPT-5.2 ships as three distinct models with separate API identifiers:
gpt-5.2-chat-latest(GPT-5.2 Instant) - fast, designed for info-seeking, how-tos, and everyday writinggpt-5.2(GPT-5.2 Thinking) - reasoning model for coding, long-document analysis, and multi-step tasksgpt-5.2-pro(GPT-5.2 Pro) - highest compute, maximum accuracy on complex tasks
All three variants default to reasoning.effort='none', a setting introduced with GPT-5.2. This is a documented departure from earlier GPT 5 models, which defaulted to medium. It lowers latency at default and gives developers explicit control to increase reasoning depth as needed.
What Separates GPT-5.2
| Trait | Detail |
|---|---|
reasoning.effort='none' as default | First GPT 5 series model to ship with reasoning off by default, making low-latency the baseline and high-effort reasoning opt-in |
/compact endpoint compatibility | Thinking variant supports the Responses API /compact endpoint, extending effective context beyond the model's physical context window for tool-heavy, long-running workflows |
| Dedicated Codex variant | Released a separate GPT-5.2-Codex model optimized specifically for agentic software engineering and defensive cybersecurity workloads |
xhigh reasoning effort | Both Thinking and Pro variants are the first to support the fifth reasoning effort level xhigh, in addition to none, low, medium, and high |
Benchmark Performance
Independent evaluations · Artificial Analysis
Accuracy & Capability Details
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | OpenAI | Anthropic | Anthropic |
| Release Date | December 11, 2025 | September 22, 2026 | September 28, 2026 |
| Knowledge Cutoff | - | - | Jun 2026 |
| Context & Limits | |||
| Context Window | 400K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | $1.75 Best Input Pricing | $4 | $2 |
| Output Pricing | $14 | $20 | $10 Best Output Pricing |
| Modalities | |||
| Inputs | fileimagetext | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 26.5 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | - | - | - |
| Agentic Index | - | - | - |
Humanity's Last Exam
Extremely difficult logical reasoning and knowledge.
Long Context Reasoning
Logical reasoning over long context windows.
SciCode Benchmark
Scientific coding and mathematical modeling.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
Explore more from OpenAI
Other models by OpenAI
Top AI Models
Leading alternatives by intelligence score
