GPT / OpenAI series comparison
OpenAI's GPT and o-series: GPT as the general flagship, the o-series focused on deep reasoning, with the most mature ecosystem.
All models
GPT / OpenAI series comparison
OpenAI's GPT and o-series: GPT as the general flagship, the o-series focused on deep reasoning, with the most mature ecosystem.
5 versions (sorted by capability)
| Version | Context | Input | Output | Reasoning | Code | Speed | Open source |
|---|---|---|---|---|---|---|---|
| GPT-4o | 128K | $5 / 1M | $15 / 1M | ●●●●○ | ●●●●○ | ●●●●○ | Closed |
| GPT-4o mini | 128K | $0.15 / 1M | $0.6 / 1M | ●●●○○ | ●●●○○ | ●●●●● | Closed |
| GPT-5.5 | — | — | — | ●●●●● | ●●●●● | ●●●●○ | Closed |
| o1-preview | 128K | $15 / 1M | $60 / 1M | ●●●●● | ●●●●● | ●●○○○ | Closed |
| o3 | 200K | — | — | ●●●●● | ●●●●● | ●●●○○ | Closed |
GPT-4oOpenAI's flagship multimodal model, proficient in vision, speech, and text, with fast response speed and the most comprehensive ecosystem.GPT-4o miniGPT-4o Lite, 3x faster than GPT-4o with 95% cost reduction, ideal for high-concurrency agent scenarios.GPT-5.5OpenAI's current flagship, ranked second on the Intelligence Index (just behind Claude Opus 4.8), with top-tier performance in high/ultra-high reasoning settings.o1-previewOpenAI's specialized reasoning model, answers after deep thinking, performs best on complex math/science/code problems, but is slow.o3OpenAI o-series reasoning model, answers after deep thinking, performs strongly on complex math/science/coding problems.