Unverified: Early Testers Praise 'Gemini 3.8' as a Well-Rounded Agentic Model
johnowhitaker · x · 2026-09-03
An unverified post claims Google's 'Gemini 3.8' is now being tested, describing it as a well-rounded agentic model that performs well across domains (finance, legal) and modalities, with impressive progress in recent months. The thread also nudges osanseviero to fix Colab's default, which still ships Gemini 2.5 Flash instead of the newer Flash 3. Model version not officially confirmed.
Related event: HF Engineer Praises Gemini 3.8 as Well-Rounded Agentic Model(3 posts)→
More from Models
- Mocked benchmark shows DeepSeek's document chart grounding badly misaligned — burkov · 2026-10-18
- Dev complains Claude's habit of saying "not done" is his biggest pain point — jasonkneen · 2026-10-18
- Open-weight TYGER speech decision model makes 80ms turn-taking calls for full-duplex voice agents — solyarisoftware · 2026-10-18
- Paying Google AI subscriber blocked: custom-domain Workspace accounts can't access AI Studio — ChrisRackauckas · 2026-10-18
- Running Ling 3.1 Flash (560B MoE, 25B Active) Across 4 DGX Sparks via llama.cpp RPC — BerryBlush_Rogers691 · 2026-10-18
- Users press OpenAI on long-delayed Adult Mode a year after it was promised — ___Patrice___ · 2026-10-18