Best Google Models for Agents

AI models ranked by the Artificial Analysis Agentic Index — measuring multi-step tool use, planning and task completion (including Tau²-Bench). The best models for building autonomous agents and agentic workflows.

1. 1G
   gemini-3.5-flash

   ReasoningToolsJSON+243.3 intel · $1.50/M · 148 t/s
   51.0

   Agentic
2. 2G
   gemini-3-flash-preview

   ReasoningToolsJSON+246.4 intel · $0.500/M · 68 t/s
   49.7

   Agentic
3. 3G
   gemini-3.1-pro-preview

   ReasoningToolsJSON+241.3 intel · $2.00/M · 85 t/s
   45.0

   Agentic
4. 4G
   gemma-4-31b-it:free

   ReasoningToolsJSON+139.2 intel · Free/M · 61 t/s
   40.9

   Agentic
5. 5G
   gemma-4-31b-it

   ReasoningToolsJSON+139.2 intel · $0.120/M · 61 t/s
   40.9

   Agentic
6. 6G
   gemini-2.5-pro

   ReasoningToolsJSON+234.6 intel · $1.25/M · 99 t/s
   32.7

   Agentic
7. 7G
   gemma-4-26b-a4b-it:free

   ReasoningToolsJSON+131.2 intel · Free/M · 40 t/s
   32.1

   Agentic
8. 8G
   gemma-4-26b-a4b-it

   ReasoningToolsJSON+131.2 intel · $0.060/M · 40 t/s
   32.1

   Agentic
9. 9G
   gemini-3.1-flash-lite-preview

   ReasoningToolsJSON+233.5 intel · $0.250/M · 108 t/s
   25.7

   Agentic
10. 10G
    gemini-2.5-flash-lite-preview-09-2025

    ReasoningToolsJSON+219.4 intel · $0.100/M · 178 t/s
    10.1

    Agentic
11. 11G
    gemini-2.5-flash-lite

    ReasoningToolsJSON+217.6 intel · $0.100/M · 103 t/s
    6.1

    Agentic
12. 12G
    gemma-3-12b-it

    ToolsJSONVision8.8 intel · $0.050/M · 30 t/s
    3.6

    Agentic
13. 13G
    gemma-3-27b-it

    ToolsJSONVision10.3 intel · $0.080/M · 42 t/s
    3.5

    Agentic
14. 14G
    gemma-3-4b-it

    JSONVision6.3 intel · $0.050/M · 19 t/s
    1.7

    Agentic