OpenAI plans to phase out GPT-4.5, its largest-ever AI model, from its API

Tech Crunch - Apr 14th, 2025

OpenAI announced plans to phase out GPT-4.5 from its API offerings by July 14, urging developers to transition to GPT-4.1, which the company claims provides similar or superior performance at a reduced cost. GPT-4.5, introduced in late February and code-named Orion, was OpenAI's most advanced model, boasting enhanced writing and persuasiveness. However, it is notably expensive to operate, with costs reaching $75 per million input tokens and $150 per million output tokens. OpenAI's decision is driven by the high operational expenses and a strategic focus on future model development.

Despite its scale, GPT-4.5 did not achieve frontier-level status on industry benchmarks, prompting OpenAI to reassess its long-term viability. The model remains accessible on ChatGPT in a research preview for paying customers but will no longer be available via API. By emphasizing GPT-4.1 as the preferred choice, OpenAI aims to meet the demand for cost-effective and efficient AI solutions while advancing its technological capabilities. This transition reflects broader trends within the AI industry, where balancing innovation with operational costs remains a critical consideration.

Ai Technology Open Ai Machine Learning Gpt 4.5 Gpt 4.1 Api Deprecation Tech Industry News

Story submitted by Fairstory

RATING

6.2

Moderately Fair

Read with skepticism

The article provides a clear and timely overview of OpenAI's decision to phase out GPT-4.5 in favor of GPT-4.1, focusing on the technical and strategic aspects of the transition. Its strengths lie in its clarity and relevance to current AI developments. However, the story could benefit from greater source transparency, a more balanced presentation of perspectives, and additional context on the implications for developers and users. While the factual claims are generally accurate, the lack of detailed source attribution and diverse viewpoints limits the depth and engagement potential of the article. Overall, it serves as an informative piece for those interested in AI technology, though it might not significantly impact broader public opinion or policy discussions.

Accuracy

The story presents several factual claims that align with known information about OpenAI's plans to phase out GPT-4.5 from its API. The timeline for the transition, the introduction of GPT-4.1 as a replacement, and the cost details for running GPT-4.5 are consistent with available data. However, the article could benefit from more precise details about the performance benchmarks where GPT-4.5 falls short and clearer comparisons with GPT-4.1. While the claims about GPT-4.5's costliness and its continued availability in ChatGPT are accurate, the lack of direct quotes or references to specific OpenAI announcements slightly reduces the verifiability.

Balance

The article primarily focuses on OpenAI's corporate decisions and technical specifications, offering limited perspectives. It provides OpenAI's rationale for transitioning from GPT-4.5 to GPT-4.1 but does not include viewpoints from developers or users affected by this change. This creates a potential imbalance, as the narrative is heavily tilted towards OpenAI's perspective without considering broader implications or criticisms from the tech community.

Clarity

The article is generally clear and concise, effectively communicating the main points about OpenAI's decision to phase out GPT-4.5. The language is straightforward, and the structure logically presents the information in a coherent manner. The use of specific terms like 'API', 'input tokens', and 'output tokens' is appropriate for the target audience, although some readers might benefit from additional context or definitions.

Source quality

The story relies on statements attributed to an OpenAI spokesperson and mentions an email to TechCrunch, suggesting some level of direct source engagement. However, the article does not provide specific names or direct quotes, which weakens the source credibility. The absence of multiple sources or corroborative evidence from independent experts or analysts further limits the article's reliability.

Transparency

The article lacks transparency in several areas, such as the absence of detailed explanations about the methodology used to evaluate GPT-4.5's performance or the criteria for selecting GPT-4.1 as a replacement. It also does not disclose any potential conflicts of interest or the basis for the claims made. Greater transparency about the source of the information and the context of OpenAI's decisions would enhance the article's credibility.