Archive — US — techcrunch
OpenAI plans to phase out GPT-4.5, its largest-ever AI model, from its API
OpenAI will phase out access to its largest AI model, GPT-4.5, from its API by July 14, 2024. Developers must switch to GPT-4.1, launched recently, which offers similar or improved performance at a lower cost. While GPT-4.5 remains available for ChatGPT users, its high operational costs influenced this decision, as OpenAI aims to focus on future developments.
Key points
- OpenAI announces the phase-out of GPT-4.5 from API access by July 14, 2024.
- Developers are encouraged to transition to GPT-4.1, launched recently as a cost-effective alternative.
- GPT-4.5, launched in late February, was OpenAI's largest model but faced operational cost challenges at $75 per million input tokens.
- Despite improvements over GPT-4, GPT-4.5 does not meet 'frontier-level' standards in industry benchmarks.
- GPT-4.5 will still be available in ChatGPT for paying customers, focusing the change on API offerings.
Why it matters
- The decision reflects OpenAI's shift towards more sustainable and cost-effective AI model deployment.
- This is in line with broader IT trends emphasizing efficiency and cost reduction for AI operations.
- OpenAI's approach highlights the growing importance of balancing model performance with operational expenses in AI development.
OpenAI's transition plan from GPT-4.5 to GPT-4.1 underscores a strategic move towards more efficient AI use, prioritizing future innovations while managing costs effectively.