OpenAI dropped the price of o3 by 80% (twitter.com)

Article summary

OpenAI has dropped the price of its o3 model by 80%. The new pricing is expected to make the model more competitive with other options on the market. The price drop was announced by Sam Altman, and the new pricing is now in effect. The o3 model is a reasoning model, and its pricing is now comparable to other models like GPT-4.1.

Main themes

• o3 price drop
• model competitiveness
• reasoning models
• naming conventions
• model performance
• quantization
• benchmarks
• client variability

What commenters say

• The 80% price drop of o3 is surprising and may be due to infrastructure improvements or competition with other models.
• The price drop may not be enough to make o3 competitive with other models like Deepseek R1, which is still significantly cheaper.
• Some users have found the bigger models like o3 and Claude 4 Opus to be disappointing for coding tasks.
• The naming conventions used by OpenAI are confusing and make it difficult to differentiate between models like o3 and GPT-4.1.
• Some users have noticed a decline in performance of models over time and suspect that this may be due to quantization.
• The performance of models can vary depending on the specific use case and the client used to access the model.
• Benchmarks that track historical performance of models are needed to determine if models are actually getting worse over time.
• Different models have different strengths and weaknesses, and what works well for one user may not work well for another.