Similar performance, security trade-offs also seen in current public models.
Read the original at Ars Technica AI: OpenAI says planned GPT-6.1 is too insecure to release
How OpenSmartRoute helps
Teams using OpenSmartRoute gain immediate value from this security shift. The router scores every candidate on safety before sending a request. Hard rules ensure text with personal data stays on an on-premises model. This prevents data leaks even if a new model is too insecure to release.
The platform learns from outcomes so a model that fails gets less traffic. An input guard spots prompt injection and personal data before a request leaves. This setup protects users without needing to wait for a new model launch.
A savings ledger shows what each routed request cost next to the most expensive option. The team sets weights per request to balance quality, cost, speed, and safety. Nothing else changes in the app when a new model arrives. A new model is just one catalogue entry competing on the next request.
Source: https://arstechnica.com/ai/2026/09/openai-says-planned-gpt-6-1-is-too-insecure-to-release/