ChatGPT vs. Telos: where to write a screenplay with AI

by Telos

To get unstuck on an idea, test ten titles or settle a formatting question, a chat AI — ChatGPT, Claude, Gemini — does the job, is flexible and costs little, and ChatGPT and Claude projects already keep files, instructions and history across chats. To write a whole screenplay, the limit is elsewhere: there, the project is reference text, not a screenplay with parts the AI reads and checks, and there's no screenplay formatting or export built in. Telos runs on Claude, by Anthropic; the difference is architectural — world, characters, structure and setups are parts of the project and enter every generation on their own.

Chat AI (ChatGPT, Claude, Gemini)Telos
PriceVaries by plan and region. Claude, for example, has a free plan and Pro at $20/month ($17/month billed annually); ChatGPT and Gemini: see their official pages$49/month + tax where applicable, with 750 credits a month for AI actions
Platforms—Web app
AIA general-purpose model, steered by what's in the conversationClaude, by Anthropic, steered by the project: writes from the world, characters, structure and voice you built, and reads the whole script, quoting the passage behind every finding
Project memoryProjects keep reference files, instructions and chat history (ChatGPT, Claude); what's there is reference textWorld, characters and arcs, structure, open setups and authorial voice are parts of the project and enter every generation automatically
Collaboration—A reading link where the studio leaves notes without creating an account (vertical drama)
Import and exportNo screenplay formatting or export built inImports Final Draft (.fdx), Fountain, PDF and plain text; exports PDF, Final Draft (.fdx) and Fountain
FormatsAny request, in free-form textFeature film, TV series, vertical drama and short film, each with its own path
Interface languages—English and Portuguese; the work can be written in English or Portuguese

What's the difference between a chat AI and Telos?

In a chat AI, you ask and the model answers based on what's in the conversation and, if you use a project, on the files and instructions you saved there. It's the most flexible tool there is: the same window handles brainstorming, a craft question and an email to a producer. For a screenplay, though, a chat project is a folder of text: you keep the bible, the outline and the character notes up to date by hand, and nothing there knows that a setup in scene 4 needs a payoff, or that a line contradicts a character's profile. And what comes out is text: formatting and exporting are on you.

Telos runs on the same kind of technology — today, Claude, by Anthropic. The difference is architectural. The project lives in a system, and the world, the characters and their arcs, the structure, every setup that hasn't paid off yet and your authorial voice enter every generation without you pasting anything. Each format has its own path: eleven phases for a feature, the season as the unit for a series, a running-time target per episode for vertical drama. The script read quotes the passage behind every finding and, where something can't be measured, says "not measured." And the editor formats the screenplay and exports to PDF, Final Draft (.fdx) and Fountain.

Telos has a conversation too: the Creation Room, for discovering world, characters, theme, voice and treatment. The difference is what happens next — when something lands, one click consolidates it into the project.

Can you write a whole screenplay in ChatGPT?

You can, and people do. The cost shows up halfway through, not at the start: in scene 4 everything is fresh; by scene 20, the character who was terrified of water is swimming, because that detail sat in a file nobody checked. The bigger the project — a feature, a season, 60 episodes of vertical drama — the more time goes into rebuilding context and checking continuity by hand.

When is a chat AI the better choice?

  • You don't have a project yet. To explore a loose premise, generate variations and throw them away without ceremony, a conversation is the right shape.
  • You have a quick question. A formatting convention, a craft term, period research: opening a chat is faster than opening a system.
  • The text isn't a screenplay. An email, a query letter, meeting notes — things that don't need your story's world.
  • Your budget is tight. Claude has a free plan, and Pro costs $20 a month ($17 billed annually) — less than Telos's $49.
  • You want total freedom of form. A chat takes any request, in any order.

When is Telos the better choice?

  • The script is long, or there are many of them. A feature, a series, a season of vertical drama: that's where holding the story matters more than writing one good scene.
  • You want AI writing in your voice. Your authorial voice is compiled from what you've written and goes into every generation.
  • You want notes backed by evidence. Every structural finding quotes the passage from your script behind it, so you can disagree on evidence instead of on authority.
  • You have to deliver. The script comes out formatted as PDF, Final Draft (.fdx) and Fountain; for a series, the season exports as a ZIP, one file per episode. For vertical drama, see our guide to writing a vertical drama.
  • You want to know the cost up front. Every AI action shows its price before you click.
  • You want to know what happens to your writing. The providers Telos calls, Anthropic and OpenAI, are bound not to train models on what goes through their API, and Telos doesn't use your writing to train anything. Price and credits on pricing.

Can you use both?

Yes, and it makes sense. Use the chat for what it does well: loose ideas and quick questions. When the idea becomes a project, bring it into Telos — it imports plain text, PDF, Fountain and Final Draft (.fdx), and character aliases merge on their own: MARIA, Maria and M. become one. If you've already drafted scenes in a chat, they come in as text and become part of the project.

From then on, what you decide about the world, the characters and the structure stays in the system, and the next generation starts out knowing it.

Sources

Prices and features checked on the official pages on Oct 16, 2026.

Written with AI assistance, from the sources cited.