Anthropic LLM¶
Sends a text prompt to Anthropic models (Claude family) through the LiteLLM provider and returns the model’s text response. Supports a system prompt, temperature control, and maximum token limits, with convenient optional inputs for templating or dynamic context. Includes built-in fallback model name mappings to help resolve common Anthropic model identifiers.
Usage¶
Use this node whenever you need a Claude (Anthropic) model to generate, summarize, translate, or reason over text. Typically, you will set the model, provide a system prompt to define behavior, pass the main user prompt (optionally enriched with input_1..input_4), and tune temperature and max_tokens. Connect its string output to downstream logic, routing, or storage nodes.
Inputs¶
| Field | Required | Type | Description | Example |
|---|---|---|---|---|
| model | True | STRING | The Anthropic model identifier to use. Accepts Anthropic Claude variants. Common examples are mapped via internal fallbacks. | claude-3-5-sonnet-20241022 |
| system_prompt | True | STRING | High-level instructions guiding the model’s behavior and style for the entire session. | You are a concise, helpful assistant. Provide clear bullet-point answers. |
| prompt | True | STRING | The main user message or task description sent to the model. You may incorporate optional inputs as context. | Summarize the following text and extract key actions: {input_1} |
| temperature | True | FLOAT | Controls randomness (0=deterministic, 1=creative). Higher values increase variability. | 0.5 |
| max_tokens | True | INT | Maximum number of tokens to generate in the response. Must be within the selected model’s limits. | 1024 |
| input_1 | False | STRING | Optional auxiliary text input for templating or adding extra context to the prompt. | Customer message: I can’t log in to my account after resetting my password. |
| input_2 | False | STRING | Optional auxiliary text input for templating or adding extra context to the prompt. | Relevant policy: Users must confirm their email after password reset. |
| input_3 | False | STRING | Optional auxiliary text input for templating or adding extra context to the prompt. | Known issue: Email delivery delays reported in region X. |
| input_4 | False | STRING | Optional auxiliary text input for templating or adding extra context to the prompt. | Support guidelines: Always verify account ownership before changes. |
Outputs¶
| Field | Type | Description | Example |
|---|---|---|---|
| Output | STRING | The generated text from the selected Anthropic model. | Here’s a concise summary with key actions and suggested next steps... |
Important Notes¶
- Model selection: Use valid Anthropic model IDs (e.g., Claude family). The node includes fallback mappings to resolve common identifiers.
- Credentials: Ensure Anthropic access is configured in your environment or project settings before use. Never paste secrets directly into prompts; store them securely as environment or project variables.
- Token limits: max_tokens must not exceed the model’s limit. If set too high, the request may fail.
- Temperature tuning: Lower values (e.g., 0.0–0.3) are better for deterministic outputs; higher values (e.g., 0.7–1.0) encourage creativity.
- Optional inputs are pass-through strings you can reference in your prompt to add context or build templates.
Troubleshooting¶
- Invalid or missing API credentials: Configure the Anthropic API key in your project or environment settings and retry.
- Model not found: Verify the model name matches an available Anthropic model or use a known fallback identifier.
- Output too long or truncated: Reduce response length by lowering max_tokens or by requesting a shorter answer in your prompt.
- Empty or low-quality responses: Provide a clearer system_prompt, add structured context via input_1..input_4, or lower temperature for more focused results.
- Rate limits or timeouts: Retry with exponential backoff, reduce request frequency, or simplify prompts to shorten processing time.