FunctionGemma

Google

Specialized Gemma 3 270M variant for native function calling and API action execution. Unified action and chat functionality - generates both structured function calls and natural language responses. 256k vocabulary optimized for JSON and multilingual tokenization. Fine-tuned to 85% accuracy on Mobile Actions evaluation. Designed for edge devices (Jetson Nano, mobile). Supports LiteRT-LM, vLLM, MLX, Llama.cpp, Ollama, LM Studio.

Text Generation Local Gemma Family v3
Parameters
270M
params
Context Window
8K
tokens
Max Output
-
tokens
Input Price
-
per 1M tokens
Output Price
-
per 1M tokens

Capabilities

👁️
Vision
Function Calling
📋
JSON Mode
🌊
Streaming
💬
System Prompt
🖥️
Code Execution
🔍
Web Search
🔌
MCP Support

Local Model Specs

Architecture
gemma-3-270m
Runtime
ollama

Details

Release Date
December 18, 2025
Knowledge Cutoff
-
Source
Local
License
gemma
Model ID
functiongemma-270m
Last updated: March 13, 2026