| Spec | Value |
|---|---|
| Maker | OpenAI[1][2] |
| Context window | 200,000 tokens |
| Maximum output | 100,000 tokens |
| Knowledge cutoff | Jun 1, 2024 |
| Input | Text, image |
| API price per 1M tokens (input / output) | $20 / $80 |
| Access | Responses API only |
From OpenAI's API model page (now Deprecated), which publishes no benchmark figures; the charts use the launch figures quoted in Patrick McGuinness's o3-pro review. · source
Estimate from the version's published specs, benchmarks and reviews, 0-5; scores are only comparable within one criteria year. Scored from OpenAI's o3-pro API model page (opened this run): 200,000-token context = 2; text and image input = 3; $20 input / $80 output per 1M tokens = $35 blended at 3:1 = 0.5 (price shown is the page's current list; McGuinness's launch review gives the same $20/$80). Reasoning, coding and speed are null: the Artificial Analysis index value and a Terminal-Bench figure for o3-pro were not opened this run (McGuinness only says AA ranks it top, with no number), and no speed was published. Independent reviews: 4 found (Latent Space hands-on, McGuinness analysis, TechCrunch and Gigazine press reports); the press ones restate OpenAI's figures. Reviewers agree it is slow and costs about 10x o3 per task on ARC-AGI for no better result.[2]
It’s a great thinker, but it’s gotta grow into being a great do-er.
OpenAI’s o3-pro is smart and slow. It gobbles tokens and extends thoughts, needing lots of context and the right use cases to show its strength in deep analysis.
O3-pro also beats Anthropic’s recently released Claude 4 Opus on GPQA Diamond, a test of PhD-level science knowledge.
It prioritizes reliability over speed
| God is hungry for Context: First thoughts on o3 pro — Latent Space · Jun 10, 2025 | 0.20 |
|---|---|
| OpenAI releases o3-pro, a souped-up version of its o3 AI reasoning model | TechCrunch · Jun 10, 2025 | 0.50 |
| OpenAI releases o3-Pro, an improved model of o3 that thinks longer and answers more accurately - GIGAZINE · Jun 11, 2025 | 0.40 |
| o3-pro, the AI That Thinks Too Much - by Patrick McGuinness · Jun 14, 2025 | -0.10 |
Photos from readers, kept apart from the pin's sourced media. Rate what you ate or saw, and on a restaurant pin tag the dishes right on your photo.
The first entry is always the pin's source. Overall confidence is a weighted average of how firmly each reference supports the start and end times used above; a reference counts half as much for every 180 days older than the newest.
OpenAI's[2] own model release notes date this entry June 10, 2025
OpenAI's[1] API model page for o3-pro says it "uses more compute to think harder", is Responses-API only, and lists a 200,000 context window, 100,000 max output tokens and $20/$80 per 1M tokens.
Reports the o3-pro release, a version of o3 for Pro users.[1]
Something missing or wrong? Say it in your own words: a link that backs this pin up, a different start or end date and why, or a fact it lacks or gets wrong. The AI checks it against this pin's sources, searches for better ones, and adds any page that backs you up. The pin's own sources still count most. A picture that shows something else, or shows it badly, is looked at too, and moved down or replaced.