According to a comprehensive report by Artificial Analysis (PDF), artificial intelligence models showed remarkable progress throughout 2024, with multiple companies catching up to and surpassing OpenAI’s GPT-4 capabilities. The report, published on artificialanalysis.ai, documents substantial improvements in model performance, efficiency, and accessibility.
The analysis reveals that frontier language models achieved new intelligence benchmarks, with models like Claude 3.5 Sonnet and Gemini 1.5 Pro demonstrating capabilities that exceed previous standards. The report indicates that open-source models, particularly those from Meta, Mistral, and Alibaba, have significantly narrowed the gap with proprietary solutions.
A notable development has been the dramatic reduction in inference pricing, with some models offering GPT-4-level performance at up to 75 times lower cost. The research shows that smaller models are now achieving intelligence levels previously only possible with much larger architectures.
Context windows have expanded significantly, with 128,000 tokens becoming the new industry standard. Some models, such as Gemini and Nova, now support up to 2 million tokens, representing a 32-fold increase since the third quarter of 2023.
The report also highlights progress in multimodal capabilities, with 61% of surveyed developers planning to implement such features. Image generation quality saw substantial improvements, particularly in photorealism and text rendering, while video generation evolved from a field dominated by OpenAI’s Sora announcement to a competitive landscape with multiple strong contenders by year’s end.
Text-to-speech and speech-to-text technologies also advanced significantly, with transformer-based models setting new quality benchmarks and dramatic improvements in processing speed and cost efficiency.