API限流 (2 articles)

GPT-o3 Crashes: 5 Rate Limits in 30 Seconds, Long Context Score Plummets by 33.5 Points

GPT-o3's long context processing capability collapsed in recent testing, with scores dropping from 62.3 to 28.8 points due to aggressive API rate limiting, exposing serious infrastructure issues at OpenAI.

GPT-4o Crashes: 5 Failed Tests Expose OpenAI's Infrastructure Crisis

GPT-4o's catastrophic failure in long-context tests, with 5 questions returning rate limit errors, reveals OpenAI's severe infrastructure problems rather than model capability issues.