AI Developer Openai Introduces GPT-4.5 as Rivalry among AI Companies Escalates, Yet Performance Issues Persist
On Thursday, OpenAI announced the launch of its latest model, GPT-4.5, to users of the ChatGPT Pro service. However, the model will not be available to other users until the following week due to limited computing capacity.
The size of the new GPT-4.5 model has not been disclosed by OpenAI. The model, internally referred to as Orion, is the last model to be built using the pre-training method employed for ChatGPT.
GPT-4.5 is one of the most anticipated products in the generative AI market. While it offers higher-functioning capabilities, it is priced 15 to 30 times higher than OpenAI's GPT-4 model.
OpenAI's blog post emphasized softer, more qualitative metrics to measure GPT-4.5's improvements over previous models. The new model is said to be less prone to "hallucinations" and demonstrates an "improved ability to follow user intent."
However, benchmark scores suggest that GPT-4.5 lags behind Anthropic's Claude 3.7 Sonnet, a rival AI model unveiled earlier this week. The key differences between the two models lie mainly in their coding performance, reasoning transparency, pricing, and use case optimizations.
Coding and Reasoning Performance
Claude 3.7 Sonnet uses “Extended Thinking Scaffolds” to enhance coding accuracy from 62.3% to 70.3% on SWE-bench and demonstrates strong structured software engineering skills and transparent chain-of-thought reasoning. It excels in business communications and structured code refactoring. GPT-4.5, while somewhat pricier, generally offers higher-functioning capabilities and additional feature sets, although exact SWE-bench scores for GPT-4.5 specifically are less detailed in the sources.
Transparency of Reasoning
Claude 3.7 Sonnet emphasizes transparent reasoning via its “extended thinking” mode, exposing error backtracking and alternative solutions during problem-solving, which is an advantage for complex analysis tasks. GPT-4.5 does not highlight this as a core feature in the same way.
Cost and Efficiency
Claude 3.7 Sonnet presents a balanced price-performance ratio with a lower cost model, making it attractive for users who need reliable, fast responses for structured tasks without high expenses. GPT-4.5 is positioned as a higher-cost yet feature-rich option, with additional capabilities that justify the price for advanced or broad use cases.
Use Case Specialization
Claude 3.7 Sonnet is optimized for business communications, structured code management, and rapid, clear responses for everyday software engineering tasks. GPT-4.5 has broader application potential, especially where multimodal inputs or additional advanced features are required, although it is less specialized in extended coding scaffolds and constructive reasoning transparency.
Release and Ecosystem Context
Although search results focus more on newer iterations like GPT-5 and Claude 4 Sonnet, Claude 3.7 Sonnet remains notable for pioneering extended thinking scaffolds that set it apart from earlier and some contemporary models, while GPT-4.5 fits between GPT-4.1 and GPT-5 in the OpenAI lineup, offering incremental improvements but apparently with less emphasis on the transparent chain-of-thought processes emphasized by Claude 3.7 Sonnet.
In summary, Claude 3.7 Sonnet excels in structured, transparent reasoning and cost-efficient coding accuracy, making it suitable for users focused on reliability and clarity in software tasks. GPT-4.5 offers a higher-priced, feature-rich platform with more general capabilities but less emphasis on the transparent stepwise problem-solving approach Claude provides. The exact numerical scores for GPT-4.5 are less specified, but the competitive context suggests GPT-4.5 is slightly behind more advanced models like GPT-5 or Claude 4 Sonnet in some benchmarks.
OpenAI has not yet decided whether it will offer GPT-4.5 as a long-term API for partners to integrate into their systems due to high operational costs. The company's CEO, Sam Altman, praised GPT-4.5, describing it as the first AI model that "felt like talking to a thoughtful person."
Technology continues to evolve with the introduction of OpenAI's GPT-4.5 model, a higher-functioning version of its previous model, priced 15 to 30 times more. However, a rival model, Anthropic's Claude 3.7 Sonnet, showcases stronger coding accuracy and transparency, offering a more cost-effective and efficient solution for structured tasks.
