Can ChatGPT make videos? Sora's shutdown and what works now

Not with Sora anymore. OpenAI discontinued the Sora web and app experiences on April 26, 2026 and the Sora API on September 24, 2026, and says no one-to-one replacement API is available. ChatGPT can still start a video by calling a video-generation tool on a remote MCP server in developer mode: the tool's service renders the clip and hands back a link.

OpenAI's side comes from its Sora discontinuation article, its video generation guide, and its ChatGPT developer mode page. Sume's side comes from the MCP overview and MCP tools and gates. All were read on 2026-09-27. If you are moving Videos API code rather than chatting, read Sora 2 API shutdown: moving video generation calls.

What happened to Sora?

Sora was OpenAI's video model, app, and API. OpenAI shut it down in two steps, and its API guide is now retained for historical reference.

• If you made videos in Sora, OpenAI's article says you can export your data, and that it will permanently delete Sora data after any final export window passes.

How do I get ChatGPT to make a video now?

Give ChatGPT a video tool. OpenAI says developer mode provides full Model Context Protocol (MCP) client support for all tools, both read and write, so ChatGPT can call a remote MCP server's video tool. With Sume's hosted MCP server, the steps are below; OpenAI's menu names were read on 2026-09-27 and may change.

• Check eligibility: developer mode is available to Pro, Plus, Business, Enterprise, and Education accounts on the web.
• Turn it on under Settings → Security and login → Developer mode.
• Go to ChatGPT Plugins, select the plus button, and create a developer-mode app for https://mcp.sume.com/mcp. How to add an MCP server to ChatGPT walks through the fields.
• Sign in on Sume's consent page and turn Write on. Read is locked on and Write is off by default; until the session has mcp:write, mutating and paid tools such as generate_video are hidden.
• In a chat, choose Developer mode from the Plus menu, select the app, and name the tool, as OpenAI suggests: "Use the Sume app's generate_video tool to make a 5-second 9:16 clip of …".

Will ChatGPT ask before it spends money?

Yes, by default. OpenAI says write actions require confirmation, that ChatGPT respects the readOnlyHint tool annotation, and that tools without it are treated as write actions. Sume's server currently marks generate_video with readOnlyHint: false, so ChatGPT asks you to approve the call. You can remember an approve or deny choice for a tool for the rest of one conversation; a new conversation prompts again.

Sume adds its own gates. generate_video is a paid generation tool, and its spend goes through your Sume wallet. dry_run=true previews the cost without submitting the job, and max_spend_usd caps a call, enforced only when provided. Every paid call also needs an idempotency_key, a stable key that deduplicates a repeated call.

How long does the video take to arrive?

Sume's video docs say generation typically takes 30 seconds to several minutes, depending on the model and parameters. In current code, generate_video answers with a job id within milliseconds, never a clip URL, and its description tells the model to wait with jobs_wait and then read jobs_result.

One jobs_wait call holds at most 55 seconds. On wait_slice_expired, the docs say to retry jobs_wait with the same id and never resubmit the paid create. Sume returns generated outputs as Sume-hosted artifacts under media.sume.com, so the finished clip's link is on that host. MCP tool call timeouts on long video jobs explains the wait loop.

What should I know before turning on developer mode?

• OpenAI labels developer mode "Elevated risk": powerful but dangerous, and intended for developers. It says to watch for prompt injections, model mistakes on write actions, and malicious MCP servers.
• Sume has no official connector for ChatGPT. This is a developer-mode app pointed at Sume's hosted MCP server.
• Sume's basics page says hosted MCP still works but is not the primary path today. For a product rather than a chat, it points to the model APIs, such as POST /v1/videos for a single clip, or to a Format for a packaged workflow.