Enterprises Flock to Open-Weight AI Alternatives for Cost Savings

Enterprises are shifting away from premium AI systems offered by Anthropic and OpenAI to open-weight alternatives in pursuit of reduced operational expenses.

According to analytics from AlphaSense, references to open AI frameworks during corporate communications increased six-fold when comparing August-September figures to the identical timeframe in the preceding year. This surge is likely due to the economic advantage offered by these open-weight frameworks, which typically demand lower operational expenditures compared to cutting-edge proprietary systems.

Telecommunications giant AT&T currently processes approximately 40% of AI operations through open models and aims to increase this proportion to 70% within the next twelve months. OpenAI has launched GPT-6 Sol and Luna variants, slashing API costs by 50% relative to previous generation pricing. Anthropic introduced Claude Opus 5.5, featuring improved token efficiency that delivers roughly 40% cost reduction compared to earlier iterations.

Robert Reilly, CFO at PNC Financial, emphasized the company's use of proprietary computing infrastructure and leveraging frontier AI systems via prominent cloud platforms while operating open-weight frameworks within company-owned data facilities.