Gemini 2.5 Flash
This node uses Google's Gemini 2.5 Flash model to generate text responses based on input prompts.
AILLMGoogleGemini
Try this nodeInputs
| Name | Type | Required | Description |
|---|---|---|---|
| instructions | string | No | System instructions for Gemini's behavior |
| input | string | Yes | The input text or question for Gemini |
| temperature | number | No | Controls randomness (0.0-2.0). Lower values are more deterministic, higher values more creative |
| maxOutputTokens | number | No | Maximum number of tokens in the response. Leave empty for model default |
| thinking_budget | number | No | Thinking budget (0-1000). Higher values enable more reasoning but increase cost and latency |
| tools | json | No | Array of tool references for function calling |
| googleSearch | boolean | No | Enable Google Search to ground responses in current web results |
| schema | schema | No | JSON schema to constrain the output format |
Outputs
| Name | Type | Description |
|---|---|---|
| text | any | Generated text or JSON response from Gemini 2.5 Flash |
| candidates | json | Full response candidates from Gemini API |
| usage_metadata | json | Token usage and cost information |
| prompt_feedback | json | Feedback about the prompt quality and safety |
| finish_reason | string | Reason why the generation finished (STOP, MAX_TOKENS, etc.) |
| grounding_metadata | json | Grounding metadata with search queries, sources, and citations from Google Search or Maps |
| tool_calls | json | Function calls made by the model |
| intermediary_messages | json | Intermediate messages and function results for debugging |
Related Nodes in Artificial Intelligence
Claude Opus 4.1
Most advanced Claude model with latest capabilities
Claude Sonnet 4
Latest Claude Sonnet model with advanced capabilities
Claude 3.7 Sonnet
Mid-tier model between 3.5 and 4.0 generations
GPT-5
Next-generation flagship model
GPT-5 Mini
Faster, cost-effective version of GPT-5
Gemini 2.5 Pro
Most capable model for complex reasoning and creative tasks
Cloudflare Model
Run any model from the Cloudflare Workers AI catalog. A searchable dropdown lists every available model; picking one rebuilds the node's inputs and outputs from its published schema.
Replicate Model
Run any model from the Replicate collection. Enter a model identifier, load its schema, and the node's inputs and outputs adapt automatically.