LLM

AIfiredcore.ai.llm

A chat / multimodal model. Reshapes its inputs, outputs, and params to the capabilities of the selected model. A provider failure fires error (with the message) instead of a reply.

Fired. It runs when an event or a value lands on trigger, and pulls its other inputs at that moment.

Inputs

Port	Type	Notes
trigger	event	Fires the node
prompt	text
tools	tool	Optional, grows: one socket per wire

Outputs

Port	Type	Notes
response	text
reasoning	text	Optional
trigger	event
error	event	Optional

Knobs

Knob	Kind	Default	Notes
model	model	none	stays a knob (cannot become an input)

Models that ship with Boltjar

The picker lists the models your connected providers offer, read live: the ones installed in Ollama and the ones your keys and endpoints reach, whether or not a manifest describes them. The ones below ship with a manifest, which adds a label, settings and what the model can do. Picking a model adds its own settings to the node as knobs, and each of those can become an input. Models explains when the list refreshes and how manifests work.

Model	Id	Needs	Can
Claude Opus 4.8	anthropic/claude-opus-4-8	ANTHROPIC_API_KEY	reads image, tools, thinking, JSON output
Claude Sonnet 4.6	anthropic/claude-sonnet-4-6	ANTHROPIC_API_KEY	reads image, tools, thinking, JSON output
Devstral 24B (Ollama)	ollama/devstral	Ollama	tools, JSON output
Gemma 4 12b (Ollama)	ollama/gemma4:12b	Ollama	reads image, tools, thinking, JSON output
Gemma 4 e2b (Ollama)	ollama/gemma4:e2b	Ollama	reads image, tools, thinking, JSON output
Gemma 4 e4b (Ollama)	ollama/gemma4:e4b	Ollama	reads image, tools, thinking, JSON output
Qwen3 14B (Ollama)	ollama/qwen3:14b	Ollama	tools, thinking, JSON output
Qwen3.5 4B (Ollama)	ollama/qwen3.5:4b	Ollama	reads image, tools, thinking, JSON output
Qwen3.5 9B (Ollama)	ollama/qwen3.5:9b	Ollama	reads image, tools, thinking, JSON output
Qwen3.5 9B q8 (Ollama)	ollama/qwen3.5:9b-q8_0	Ollama	reads image, tools, thinking, JSON output
GPT-4.1	openai/gpt-4.1	OPENAI_API_KEY	reads image, tools, JSON output
GPT-4.1 mini	openai/gpt-4.1-mini	OPENAI_API_KEY	reads image, tools, JSON output
GPT-4o	openai/gpt-4o	OPENAI_API_KEY	reads image, tools, JSON output
GPT-4o mini	openai/gpt-4o-mini	OPENAI_API_KEY	reads image, tools, JSON output
GPT-5	openai/gpt-5	OPENAI_API_KEY	reads image, tools, JSON output
GPT-5 mini	openai/gpt-5-mini	OPENAI_API_KEY	reads image, tools, JSON output
GPT-5 nano	openai/gpt-5-nano	OPENAI_API_KEY	reads image, tools, JSON output
Grok 4.20 (xAI, non-reasoning)	xai/grok-4.20	XAI_API_KEY	reads image, tools, JSON output
Grok 4.3 (xAI)	xai/grok-4.3	XAI_API_KEY	reads image, tools, thinking, JSON output

Example

Pick a model and the node reshapes: a model that reads images grows an image input, and the model’s own settings (temperature, top_p and the rest) show up as knobs you can convert to inputs. No model is picked for you (Models). With none picked, or a cloud model whose key is missing, it answers with the free offline mock, so a graph runs before you add any key. When the provider fails, error fires with the message and nothing else does.

Wire Tool nodes into tools to let a model with tool support call them. Tool calls work with Ollama, xAI, Anthropic, OpenAI and other OpenAI-compatible models, up to six rounds of calls per reply. The tool-calling agent guide builds one.

More in AI

• Embed: Turn text into an embedding vector with the embed model you pick, the way the LLM picks its model.
• Rerank: Reorder candidates by how well each one answers the query, and keep the best.
• STT: Speech to text.
• Tool: A tool the LLM can call.
• Tool Args: Where a tool’s body starts.
• TTS: Text to speech.