AI Models Comparison
This overview helps you find the optimal AI model for your requirements. The table shows key metrics such as costs, token limits, and hosting options.
AI Models Comparison
Model
| Provider | Input Cost per 1M tokens | Cached Input Cost per 1M tokens | Output Cost per 1M tokens | Input Token Limit | Output Token Limit | Batch API | Hoster | Training Cut-off | Total Cost 1M Input + 0.5M Output |
|---|
| o1 |
| gpt 4.1 |
| GPT-4o |
| gpt 4.1-mini |
| gpt 4.1-nano |
| o3-mini |
| Claude Sonnet 3.7 |
| Claude Haiku 3.5 |
| Deepseek R1 |
| Deepseek V3 |
| GPT-4o mini |
| Gemini 2.5 Pro |
| Gemini 2.0 Flash |
| Gemini 2.0 Flash-Lite |
| OpenAI | 15$ | 7.5$ | 60$ | 200,000 | 100,000 | 50% for 24 hours | OpenAI, Azure | October 2023 | 45$ |
| OpenAI | 2$ | 0.5$ | 8$ | 1,000,000 | 32,768 | 50% for 24 hours | OpenAI, Azure | June 2024 | 6$ |
| OpenAI | 2.5$ | 1.25$ | 10$ | 128,000 | 16,384 | 50% for 24 hours | OpenAI, Azure | October 2023 | 7.5$ |
| OpenAI | 0.4$ | 0.1$ | 1.6$ | 1,000,000 | 32,768 | 50% for 24 hours | OpenAI, Azure | June 2024 | 1.2$ |
| OpenAI | 0.1$ | 0.025$ | 0.4$ | 1,000,000 | 32,768 | 50% for 24 hours | OpenAI, Azure | June 2024 | 0.3$ |
| OpenAI | 1.1$ | 0.55$ | 4.4$ | 200,000 | 100,000 | 50% for 24 hours | OpenAI, Azure | October 2023 | 3.3$ |
| Antrophic | 3$ | Up to 90% | 15$ | 200,000 | 8,192 | 50% for 24 hours | Google, Amazon, Antrophic | April 2024 | 10.5$ |
| Antrophic | 0.8$ | Up to 90% | 4$ | 200,000 | 8,192 | 50% for 24 hours | Google, Amazon, Antrophic | July 2024 | 2.8$ |
| Deepseek | 0.55$ | 0.14$ | 2.19$ | 64,000 | 8,192 | Deepseek | July 2024 | 1.645$ | |
| Deepseek | 0.27$ | 0.07$ | 1.1$ | 64,000 | 8,192 | Deepseek | December 2024 | 0.82$ | |
| OpenAI | 0.15$ | 0.075$ | 0.6$ | 128,000 | 16,384 | 50% for 24 hours | OpenAI, Azure | October 2023 | 0.45$ |
| 1.25$ | 10$ | 1,048,576 | 65,536 | January 2025 | 6.25$ | ||||
| 0.1$ | 0.025$ | 0.4$ | 1,048,576 | 8,192 | June 2024 | 0.3$ | |||
| 0.075$ | 0.01875$ | 0.3$ | 1,048,576 | 8,192 | June 2024 | 0.225$ |
All costs are listed in US dollars per 1 million tokens. When selecting a model, consider not only the costs but also the training data cutoff date and maximum token limits for your specific use case.