ClaudeNode - ComfyUI Built-in Node Documentation

Generate text responses from Anthropic’s Claude models. Provide a text prompt and optionally one or more images for multimodal context, and the node returns the model’s generated text response.

Inputs

Inputs are grouped into common settings, model-specific settings that appear when a model is selected, and optional reference images.

Common Inputs

Parameter	Description	Data Type	Required	Range
`model`	The Claude model used to generate the response. Selecting a model reveals model-specific settings below.	DYNAMIC_COMBO	Yes	`"Opus 5"` `"Opus 4.8"` `"Fable 5"` `"Sonnet 5"` `"Opus 4.7"` `"Opus 4.6"` `"Sonnet 4.6"` `"Sonnet 4.5"` `"Haiku 4.5"`
`prompt`	Text input to the model. (default: empty string)	STRING	Yes	N/A
`seed`	Seed controls whether the node should re-run; results are non-deterministic regardless of seed. (default: 0)	INT	Yes	0 to 2147483647
`system_prompt`	Foundational instructions that dictate the model’s behavior. (default: empty string)	STRING	No	N/A

Opus 5 and Fable 5 Inputs

These two models share the same settings. They do not expose a temperature setting, and reasoning is always enabled.

Parameter	Description	Data Type	Required	Range
`max_tokens`	Maximum number of tokens to generate (includes reasoning tokens when enabled). (default: 32768)	INT	Yes	4096 to 64000
`reasoning_effort`	Extended thinking effort. Reasoning is always enabled for this model. (default: “high”)	COMBO	Yes	`"low"` `"medium"` `"high"`

Opus 4.8 and Sonnet 5 Inputs

These two models share the same settings. They do not expose a temperature setting.

Parameter	Description	Data Type	Required	Range
`max_tokens`	Maximum number of tokens to generate (includes reasoning tokens when enabled). (default: 32768)	INT	Yes	4096 to 64000
`reasoning_effort`	Extended thinking effort. `"off"` disables reasoning. (default: “off”)	COMBO	Yes	`"off"` `"low"` `"medium"` `"high"`

Opus 4.7, Opus 4.6, Sonnet 4.6, and Sonnet 4.5 Inputs

These four models share the same settings.

Parameter	Description	Data Type	Required	Range
`max_tokens`	Maximum number of tokens to generate (includes reasoning tokens when enabled). (default: 32768)	INT	Yes	4096 to 64000
`temperature`	Controls randomness. 0.0 is deterministic, 1.0 is most random. Ignored for Opus 4.7 and any model when `reasoning_effort` is set. (default: 1.0)	FLOAT	Yes	0.0 to 1.0 (step: 0.01)
`reasoning_effort`	Extended thinking effort. `"off"` disables reasoning. (default: “off”)	COMBO	Yes	`"off"` `"low"` `"medium"` `"high"`

Haiku 4.5 Inputs

This model does not expose a reasoning_effort setting.

Parameter	Description	Data Type	Required	Range
`max_tokens`	Maximum number of tokens to generate (includes reasoning tokens when enabled). (default: 32768)	INT	Yes	4096 to 64000
`temperature`	Controls randomness. 0.0 is deterministic, 1.0 is most random. Ignored for Opus 4.7 and any model when `reasoning_effort` is set. (default: 1.0)	FLOAT	Yes	0.0 to 1.0 (step: 0.01)

Reference Inputs

Parameter	Description	Data Type	Required	Range
`images`	Optional image(s) to use as context for the model. Up to 20 images. Growable slot: connect 1 to 20 items (`image_1` … `image_20`).	IMAGE	No	0 to 20 images

Parameter Constraints

Image limit: A maximum of 20 images can be provided per request. Connecting more than 20 images raises an error.
Prompt required: The prompt must contain at least one non-whitespace character. An empty prompt raises a validation error.
Temperature handling: When thinking is enabled, the Anthropic API requires temperature to be unset (it defaults to 1.0). Opus 5, Opus 4.8, Fable 5, and Sonnet 5 do not expose a temperature setting. Opus 4.7 ignores temperature, and any model with reasoning_effort set to "low", "medium", or "high" also ignores it.
Reasoning/thinking behavior: The reasoning_effort setting controls whether thinking is enabled. Opus 5 and Fable 5 always have reasoning enabled. Haiku 4.5 does not support reasoning. When thinking is enabled, the node uses the appropriate thinking mode for the selected model, either adaptive or budget-based. In budget mode, the reasoning token budget is capped to leave at least 1024 tokens for the actual response.
Safety refusal: If Claude declines to answer the request for safety reasons, the node raises an error asking you to rephrase the prompt or try a different model.
Output text: Thinking and reasoning blocks are not included in the output; only generated text is returned.

Outputs

Output Name	Description	Data Type
`output`	The generated text response from the Claude model. Thinking/reasoning blocks are not included. If no text is generated, returns “Empty response from Claude model.”	STRING

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): b0381e7981e5886d66b6976c7ddcad3f142bdd803271a6ac8567293dcddaa98a

​Inputs

​Common Inputs

​Opus 5 and Fable 5 Inputs

​Opus 4.8 and Sonnet 5 Inputs

​Opus 4.7, Opus 4.6, Sonnet 4.6, and Sonnet 4.5 Inputs

​Haiku 4.5 Inputs

​Reference Inputs

​Parameter Constraints

​Outputs