Google has launched Gemini Decision 1, a new multimodal decision model designed for single-pass classification tasks. This model can process up to 1,000 questions per request, with a latency of approximately 200ms and a cost of $0.10 per million input tokens, making it significantly cheaper than previous models like Flash-Lite. While Google's internal benchmarks show Gemini Decision 1 outperforming competitors like Jev 1.14 on several metrics, independent evaluations present a more mixed picture, with some benchmarks showing it tied with or behind other models in terms of overall score, cost, and latency. AI
IMPACT This release offers a potentially cheaper and faster alternative for classification tasks, which could influence how developers build agentic systems.
RANK_REASON Google's announcement of a new model, Gemini Decision 1, with specific capabilities and pricing. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
- Anthropic
- Cloudflare
- Drex Max v1.1
- Elephant 9B
- Flash-Lite Model
- Gemini Decision 1
- GLiDE v1.2
- Jev 1.14
- Logan Kilpatrick
- Michelle M. Chen
- SerpApi
- Sundar Pichai
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →