LLM Releases, Token Limits, and Pricing Since GPT 3.5 (www.metemorris.com)

🤖 AI Summary
Since the launch of ChatGPT powered by GPT-3.5 in November 2022, the landscape of large language models (LLMs) has dramatically evolved, particularly in terms of model releases and performance capabilities. The frequency of flagship model launches from major players like OpenAI, Anthropic, and Google has decreased, with median release times improving from 119 days in 2023 to 47.5 days by 2025, although rising again to 57 days in 2026. Notably, while flagship models have seen significant increases in token limits—output capacity growing from approximately 4K tokens in 2023 to 128K in 2026—economy models have become less frequent, raising questions about the sustainability of various market tiers. This rapid evolution in LLMs is crucial for the AI/ML community as it reflects the growing demands for more powerful and efficient frameworks. For developers and businesses, updates in input and output token limits and pricing strategies indicate shifting norms in model usage and cost structures. Notably, total output and pricing ratios have showcased variability; flagship input prices have slumped since early launches but experienced fluctuations, underscoring the competitive landscape spurred by open-weight models. As a result, the community must adapt to both the technical advancements in LLMs and the implications of their evolving economic models.
Loading comments...
loading comments...