Google · released 2025-11
Gemini Nano 2
On-device first. Free inference, private by default.
context
32K
output max
4K
input /M
Free
output /M
Free
modalities
T
Verdict
What it's actually good at
Strengths
- +Free on-device inference
- +Private by design
- +Zero latency to server
Weaknesses
- −Tiny context
- −Limited capability ceiling
Our scores
reasoning
5.0
coding
4.0
writing
6.0
speed
10.0
value
10.0
Nano runs on-device on Chrome and Pixel. Free inference, private by default. Best for lightweight UX features.
Pricing
Cost at common volumes
| Monthly volume | Input | Output | Est. monthly |
|---|---|---|---|
| Side project | 1M tok | 0.3M tok | $0.00 |
| Growing app | 50M tok | 15M tok | $0.00 |
| Production | 500M tok | 150M tok | $0.00 |
| Scale | 5000M tok | 1500M tok | $0.00 |
Estimates assume uncached input. Prompt caching and batch APIs can cut this by 50–90% for many workloads. Use the calculator →
Head-to-head