Largest open Gemma 3 instruction model for multilingual text generation and visual understanding
Specification
most-agreed values
Context
131K
Max output
16K
Released
2025-03-12
Knowledge cutoff
2024-08
Retires
—
Open weights
yes
Input
text, image
Output
text
Price
US dollars per million tokens · most-agreed
Input
free
Output
free
Quality
4 benchmarks · 2 sources
Composite
28th
percentile of 344 scored models
Rank
248 / 344
±0.06 sd
Evidence
4 × 2
3 or more benchmarks, from 2 or more sources
Effort range
—
one configuration only
coding5th1/3 bench
reasoning27th1/4 bench
math47th2/7 bench
Every score
one row per source and configuration — nothing averaged away
| Benchmark | Score | Configuration | Source | Run |
|---|---|---|---|---|
| MATH level 5 | 74.0 ±1.1 | default | Epoch AI | 2025-03-13 |
| GPQA diamond | 47.7 ±2.7 | default | Epoch AI | 2026-08-27 |
| OTIS Mock AIME 2024-2025 | 22.5 ±5.3 | default | Epoch AI | 2026-08-27 |
| Aider polyglot | 4.9 | default · aider whole | Aider | 2025-03-15 |
The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.
Available from 1 host
| Host | In $/M | Out $/M | Cache rd | Cache wr | Context | Output | Retires |
|---|---|---|---|---|---|---|---|
| empiriolabs | free | free | — | — | 131K | 16K | — |
More from Google
most-hosted first