Gemini 3 Flash
Model ● Covered in 4 stories + Follow
Gemini 3 Flash is Google's mid-tier language model designed for speed and efficiency, released in late 2025 as a faster and cheaper alternative to Gemini 3. The model processes over 1 trillion tokens per day on Google's API, achieves 90.4% accuracy on GPQA Diamond benchmarks, and costs $0.50 per million input tokens while running 3 times faster than its predecessor. Recent evaluations show Gemini 3 Flash exhibits fewer cascading failures compared to open-source models in enterprise IT automation tasks, though a subsequent price increase to $1.50 per million tokens has diminished its cost advantage positioning.
Updated 7 August 2026
Specifications
No specifications recorded yet.
Latest developments
Gemini Flash Gets Pricey, AI Act Delays, Agents Drive Online Traffic
The Batch ·
31
IBM and UC Berkeley Diagnose Why Enterprise Agents Fail Using IT-Bench and MAST
Hugging Face · 6 months ago ·
44
Google's year in review: 8 areas with research breakthroughs in 2025
Google DeepMind · 8 months ago ·
40
Gemini 3 Flash: frontier intelligence built for speed
Google DeepMind · 9 months ago ·
20
2026
- Gemini Flash Gets Pricey, AI Act Delays, Agents Drive Online Traffic
- IBM and UC Berkeley Diagnose Why Enterprise Agents Fail Using IT-Bench and MAST