One click, whole workflow
Recipes are here. Package a whole workflow and run it in one click
Explore recipesChatGPT
byOpenAI
The thinking layer of your canvas: expand a one-line idea into a treatment, read a reference image, and hand clean prompts to every node downstream
How ChatGPT works
You choose a tier, set a voice once, attach a reference if you have one, and ask for a defined deliverable. The reply is plain text, ready to drop into the next node's prompt box.
Pick a tier for the job
Choose fast and cheap for brainstorming, or a deeper tier for treatments, long scripts and dense image critique.
Set the voice, attach references
Write your role, voice and output format once in the system prompt, then attach images you want it to read.
Send and pipe the text on
Ask for a specific deliverable, then wire the plain-text response into image, video or audio nodes.
What ChatGPT is good at
Fluent, organized writing that takes direction: it reads images, follows camera and lighting vocabulary, holds a brand voice, and spans fast drafting tiers to deep-reasoning ones.
Reads your reference images
Attach a poster, layout or screenshot and get notes on type hierarchy, palette and composition. Say what to look at, and the answer stays on target. Images and text attach; video and audio do not.
Moodboard to shot list
Hand it a rough idea or a reference frame and get a structured breakdown: shot number, framing, lens, lighting, action. Newer tiers follow camera and lighting vocabulary literally.
Prompts for downstream nodes
Describe the look you want and receive prompts written for image and video models. Use a Luna, mini or nano tier to batch dozens of variations, then a Sol or Astra tier for the hero version.
One brand voice, held steady
Lock tone in the system prompt and every line of copy follows it. The default route covers everyday drafting and structured output; an alternate route stands by as a fallback if the default is throttled.
Made with ChatGPT
A gallery of creative instructions and what comes back: critiques, shot lists, briefs, voiceover and song direction, each written to be handed to a person or another node.
Design critique that ends in a recreate-this-look prompt
A logline unfolded into a parseable shot list
Messy notes become a structured art direction brief
Scriptwriting with a defined, strange voice
Sonic direction and lyric draft in one pass
What people build with ChatGPT
Art directors, designers, filmmakers, prompt writers and copywriters reach for it when a canvas needs a thinking layer that reads references and writes in a defined voice.
ChatGPT specs
Plans and credits- Versions
- gpt-6.1-sol, gpt-6-astra, gpt-6-luna, gpt-6-sol, gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna, gpt-5.5, gpt-5.4, gpt-5.2, gpt-5.1, gpt-5, gpt-5-mini, gpt-5-nano, gpt-4.1, gpt-4.1-mini, gpt-4.1-nano, gpt-4o, gpt-4o-mini, o1, o3-mini, o3, o4-mini
- Inputs
- Prompt (text), System Prompt (text)
- Output
- text
ChatGPT node questions
Yes. The default gpt-6.1-sol has a context window of about 1.05 million tokens, so a full screenplay or brand book fits with room for your instructions. Attach it as text or paste it into the prompt, then say exactly which parts to use, such as tone rules or character names. Ask for one scene or section per run when you need detailed output.
Fuser runs the GPT-6 and GPT-5.5/5.6 tiers, including the default gpt-6.1-sol, with reasoning on, and OpenAI doesn't accept Temperature or Top P while these models reason. Those settings are left out for these tiers, so steer variety with instructions instead, like "give five very different takes". Older tiers such as gpt-4.1 and gpt-4o still respond to Temperature.
Use Luna, mini or nano tiers for quick brainstorming and batches of prompt variations, and Sol or Astra tiers for multi-step work like treatments with continuity, long scripts and dense image critique. As reported, Astra is the deepest, Sol is the near-flagship balance and Luna is fast and cheap. The default gpt-6.1-sol also handles very long pasted documents like a full brand book or screenplay, so there's no need to drop to an older tier like GPT-4.1 for length. Tier names and availability shift quickly, so treat the dropdown as a menu of speed, depth and cost.
No. Only text and images attach to this node, so it can't watch video or listen to audio. It can read still images, such as posters, layouts and screenshots, for composition, palette and type. It also can't generate images, audio or video itself, but it can write the prompts for nodes that do.
Ask for plain prose with no markdown or preamble, and request only the prompt rather than a prompt plus explanation. Define role and format in the system prompt, and set a stop delimiter like ### to end after one prompt. Turn on Last Message Only for independent one-shot calls so earlier turns don't bleed in.
The models come from OpenAI. The lineup spans releases from May 2024 to September 2026, from GPT-4o and the o-series to the GPT-5.x and GPT-6 tiers. The node's default is gpt-6.1-sol.
Try ChatGPT on Fuser
The thinking layer of your canvas: expand a one-line idea into a treatment, read a reference image, and hand clean prompts to every node downstream
Recipes using ChatGPT
Explore all recipes
Jean Phil Style AI Character Generator@hirad
Enter a character idea—or leave SURPRISE ME—to create three character sheets and written profiles for social content.
Social & Content
View 
Prompt Extractor@maggie
Upload a reference image to receive a detailed, structured text prompt describing how to recreate its visual characteristics.
Art & Illustration
View 
Render Like a Reference Style@maggie
Combine a source image with a render-style reference to create a generated image inspired by the reference’s materials, lighting, and atmosphere.
Art & Illustration
View 
Re-Compose + Frame Your Image@laure
Upload one image to explore golden spiral, rule-of-thirds, and horizontal or vertical symmetry compositions.
Photography
View 
Translate to Voice Video (Male)@maggie
Provide a scenario image, script, voice, and language to create a talking character video with translated speech.
Social & Content
View 
Translate to Voice Video (Female)@maggie
Turn a script, scenario image, voice selection, and target language into a talking character video.
Social & Content
View Used for
Creative agencies
Develop pitches, campaign assets and interactive client work in Fuser. Combine AI generation, editable design and reusable processes across your agency.
Industry
View 
Brand identity & systems
Explore a brand’s visual identity, refine typography and build real applications. Combine AI imagery with editable design and reusable workflows in Fuser.
Use case
View 
Architecture & interiors
Explore interior moods, material palettes and architectural concepts. Turn room references and sketches into clear client presentation boards in Fuser.
Industry
View 
Film & animation
Develop film treatments, character and environment concepts, storyboards and motion studies. Connect image, video and 3D exploration in Fuser.
Industry
View 
Concepting & moodboards
Turn a creative brief into distinct visual concepts, moodboards and pitch layouts. Explore references, imagery and motion together in Fuser.
Use case
View 
Social content
Create social graphics, carousel artwork and short videos with a recognizable brand style. Keep copy editable and recurring content easier to produce in Fuser.
Use case
View 
Video & storyboards
Develop storyboard frames, shot variations and video studies from your references. Compose clear boards and explore movement in one Fuser project.
Use case
View 