@trevin says he ran some quick tests of the new Gemini Decision 1 model in a project where he already uses @typesafeai Jev and Microsoft Decision 1. Gemini Decision 1 was accurate but never cheaper, and worse in his "request matching" tests.
- 522 views
- 2 likes
- 1 bookmarks
- 0 reposts
I ran some quick tests for this new Gemini Decision 1 model in a project I'm working on where I'm already using @typesafeai Jev and Microsoft Decision 1. Gemini…
About this case
Benchmark. Trevin Chow (@trevin) says the tests were quick, in a project he is working on. Gemini Decision 1 was accurate, but it was never cheaper, and it was worse in his "request matching" tests. Photo attached. He quotes Logan Kilpatrick (@OfficialLoganK), who introduces Gemini Decision 1 as a frontier decision model in the Gemini API, at $0.10 per 1M input tokens with no additional cost, with native support for image, audio, video and PDF, a 256K context window, and 1K questions, and says it can be tried for free in the API.
Original posthttps://x.com/trevin/status/2120627626666787116