GPT-6.2 Astra
OpenAI's frontier model for professional work: GPT-6.2 Astra matches slide templates, builds finance-grade workbooks, and runs scientific work in a terminal.
Model Overview
Capabilities, design details, and architectural traits
GPT-6.2 Astra - a new quality bar for complex professional work
GPT-6.2 Astra is OpenAI's frontier model for the hardest projects, released alongside GPT-6.2 Sol, the efficient model for everyday work. Its defining idea is professional output quality: slides, spreadsheets, and documents that read less like AI and more closely follow the templates you provide and the conventions of your field. It also gets better results with fewer output tokens, so users accomplish more within their usage limits.
| Trait | What sets it apart |
|---|---|
| Template-faithful slides | Matches a slide template's layout, colors, and typography down to the exact font size, from an original file or even a screenshot, and carries the template's writing style into new slides. |
| Field-convention documents | Trained to follow the conventions professionals use in law, sales, finance, and accounting, with more direct, specific writing and less filler. |
| Built for scientists | OpenAI calls it the best model in the world for scientists. |
| First at the AI Self-Improvement threshold | The first externally released OpenAI model to meet the High threshold for AI Self-Improvement under the Preparedness Framework. |
| Astra for Law | Has a dedicated configuration for lawyers. |
| Frontend and code quality | Writes cleaner code and builds frontends that look and work better, particularly for websites, gaming, and 3D design. |
Outputs that lead with the decision
The model's work products are organized around what a reader needs. In data-science memos it leads with the result leadership needs for a rollout decision, and its slide titles state the main takeaway directly. In financial workbooks it places the revenue mix and the valuation bridge beside the sensitivity table, so readers can trace how each figure feeds the final estimate.
Reasoning built for long-horizon work
Like OpenAI's other reasoning models, GPT-6.2 Astra is trained through reinforcement learning to think before answering, producing a long internal chain of thought where it refines its approach, tries different strategies, and recognizes mistakes. This underpins its performance on long-horizon professional workflows.
Benchmark Performance
Independent evaluations · Artificial Analysis
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | OpenAI | Anthropic | |
| Release Date | November 17, 2026 | November 9, 2026 | November 18, 2026 |
| Knowledge Cutoff | - | - | Mar 2026 |
| Context & Limits | |||
| Context Window | 1.1M Best Context Window | 1M | 1.0M |
| Pricing (per 1M tokens) | |||
| Input Pricing | $8 | $8 | $3 Best Input Pricing |
| Output Pricing | $40 | $40 | $15 Best Output Pricing |
| Modalities | |||
| Inputs | textimagefile | textimagefile | textimagefileaudiovideo |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 62.1 | 64.6 Best Intelligence Index | 61.4 |
| Coding Index | - | - | - |
| Agentic Index | - | - | - |
Benchmark comparison not available
Select at least two models that have available benchmark evaluation data to render the performance comparison charts.
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