The term 'GPT-5 Class' is misleading because the frontier of AI advances so rapidly. In this 2026 scenario, the original GPT-5 is already deprecated and scores below the median, not because it degraded, but because benchmarks got harder and the entire field advanced. True comparison requires looking at the current top tier, not a stale version number.
A large majority of performance benchmarks for open-source models are self-reported by vendors, not independently verified. Therefore, claims of surpassing a proprietary model like GPT-5 should be treated as a starting hypothesis to be tested with your own data, rather than an established fact to be built upon.
The choice to self-host isn't about a 'free' model versus a paid API. It's a trade-off between a variable per-token bill and a massive fixed GPU bill plus operational overhead. Self-hosting only becomes economical when you have enough consistent workload to keep the expensive hardware perpetually busy; otherwise, an API is cheaper.
A production AI agent performs tasks of varying difficulty. Forcing all requests through a single, expensive frontier model is inefficient. A better architecture routes tasks to the most appropriate model: small, cheap open models for high-volume, low-difficulty work like retrieval, reserving the costly frontier API only for high-stakes reasoning where it matters.