---
title: "Pika's Enact lets creators direct AI video mid-scene, with answers in under a second"
description: "Enact, posted November 18, lets a creator drag an AI actor mid-scene, talk to the cast by name and redo the camera after locking a performance. Pika open-sourced the model underneath and kept the directing layer closed, a bet that value in AI video is moving from model quality to workflow."
publisher: "The Inference"
section: "Model"
published: 2026-11-20T18:20:45.641Z
modified: 2026-11-20T18:20:45.641Z
canonical: https://theinference.org/article/pika-s-enact-lets-creators-direct-ai-video-mid-scene-with-answers-in-under-a-second
language: en
keywords: "AI Video, Pika, Generative Video, Open Weights, Model Releases"
---

# Pika's Enact lets creators direct AI video mid-scene, with answers in under a second

> Enact, posted November 18, lets a creator drag an AI actor mid-scene, talk to the cast by name and redo the camera after locking a performance. Pika open-sourced the model underneath and kept the directing layer closed, a bet that value in AI video is moving from model quality to workflow.

[Pika](https://theinference.org/markets/companies/pika) posted Enact on November 18: a research preview where you direct a scene instead of describing it and waiting for a render [1][2]. Drag a character with the mouse and the scene keeps playing around the move; give a spoken note to one actor by name; lock the performance, then redo the camera [2][3]. [Pika](https://pika.art/pika-h3-flex), an AI video company backed by the venture firm [Lightspeed](https://theinference.org/markets/companies/lightspeed) [4], published the underlying model's weights for anyone to download in late October and is keeping the directing layer closed.

## Streaming speed is what makes direction possible

Enact runs on [H3 Flex, Pika's post-trained version of MiniMax's H3 video model](https://pika.art/), a base built elsewhere and then tuned by Pika's researchers for speed and control [1][4]. They cut generation to six denoising steps, the passes a model runs to turn noise into a picture, from the original's 50 [1][4]. A model that finishes a clip in seconds can absorb a note while the scene is still playing; one that takes minutes cannot. That speed is the whole trick.

Enact answers input in under a second [2]. It streams continuously rather than delivering finished chunks, and it was trained on synthetic scenes that a slower, higher-quality model filmed photoreal [2][3]. An agent layer keeps each note on the right character, so a direction to Sarah changes Sarah, not Darren [3]. Chief executive Demi Guo names the cost honestly: the responsive model is slightly lower in image quality than the offline one [3].

![Frame from Pika's Enact preview: a close-up of a woman in glasses, subtitled "Yeah, we're ready" over the name tag "ALICE"](https://storage.googleapis.com/theinference-public/img/a24e5a46095aed72.jpg)

*A frame from Pika's Enact preview, posted November 18: the actor's name tag and a spoken line sit over the generated scene, the layer a creator talks to while it plays. (Pika via X)*

## The model is free, the workflow is the product

One layer up sits the money. Pika published H3 [Flex](https://theinference.org/markets/companies/flex)'s weights on [Hugging Face](https://theinference.org/markets/companies/hugging-face) and GitHub [when the model launched on October 29](https://www.businesswire.com/news/home/20261029115507/en/Pika-Launches-Pika-H3-Flex-a-Video-Model-Built-for-Professional-Iteration) [4], and its platform already [sells rivals' models](https://theinference.org/article/minimax-s-partners-now-sell-its-video-model-for-half-the-lab-s-own-price), Kling 4.0 and Seedance 2.5 among them, charged by the credit [1][5]. The open release stops short of Pika's hosted upsampler and serving stack [5], and nothing in [the Enact announcement](https://www.vp-land.com/stories/pika-introduces-enact-a-research-preview-for-directing-ai-actors-and-the-camera-in-real-ti) says the directing layer is open either [6].

Quality, meanwhile, is bunching. On [the leaderboard Pika itself publishes](https://pika.art/pika-h3-flex), H3 Flex sits two places above the MiniMax base it came from, and it is the highest-ranked model with open weights on that board [5].

*Chart: **The video quality board is bunched, and Pika sits seventh** Bar chart of DesignArena text-to-video quality scores with Pika H3 Flex highlighted mid-table, close to the MiniMax base it was post-trained from*

*DesignArena text-to-video quality scores as published on Pika's own H3 Flex page, November 2026; higher is better. [5]*

The rivals hold the adjacent ground. Runway, another AI video company, sells editing first: [Aleph 2.0 in Edit Studio previews a change as an image before committing a generation](https://runway.com/news/introducing-aleph-2-and-edit-studio) [7]. Luma, a third, pitches Ray 3.2 as directing video the way you would direct a shoot [8]. Enact goes further, because the direction lands while the scene is still unfolding [1][3].

## Shipping, not the demo, decides what Pika is worth

Today Enact is a demo on request, with no pricing, release date or public access [6]. Pika is private, so what moves here is private-market positioning rather than a listed price.

The preview still has to prove latency on real inputs, how long a scene can run and how many characters can take a note at once [6]. If the directing layer ships into the paid platform, Pika would be the only one of the three taking direction mid-performance rather than between renders [1][7][8].

The model layer is now free to download. The directing layer is not, and turning it from a demo into a tool working directors finish scenes in is the work that decides what Pika is worth.

## Takeaway

Pika is giving away the model and keeping the directing layer: the bet is that in AI video the workflow, not the render, is what sells.

## Sources

1. [Pika homepage, Research Preview: Pika Enact listing and H3 Flex research notes, 18 November 2026](https://pika.art/)
2. [TLDR, Pika Labs releases Pika Enact live directing environment, 18 November 2026](https://tldr.tech/story/1351c8db-2d44-4fb6-8e06-8a7427e930a1)
3. [The Neuron, WATCH: Pika lets you direct AI actors live, 19 November 2026](https://www.theneurondaily.com/p/watch-pika-lets-you-direct-ai-actors-live)
4. [Business Wire, Pika Launches Pika H3 Flex, a Video Model Built for Professional Iteration, 29 October 2026](https://www.businesswire.com/news/home/20261029115507/en/Pika-Launches-Pika-H3-Flex-a-Video-Model-Built-for-Professional-Iteration)
5. [Pika, H3 Flex product page: leaderboard, open-weights scope and pricing, accessed 20 November 2026](https://pika.art/pika-h3-flex)
6. [VP Land, Pika Introduces Enact, a Research Preview for Directing AI Actors and the Camera in Real Time, 19 November 2026](https://www.vp-land.com/stories/pika-introduces-enact-a-research-preview-for-directing-ai-actors-and-the-camera-in-real-ti)
7. [Runway, Introducing Aleph 2.0 and Edit Studio, 21 May 2026](https://runway.com/news/introducing-aleph-2-and-edit-studio)
8. [Luma homepage, Ray 3.2: Direct Every Frame, accessed 20 November 2026](https://lumalabs.ai/)


---

Pika's Enact lets creators direct AI video mid-scene, with answers in under a second — The Inference. Canonical: https://theinference.org/article/pika-s-enact-lets-creators-direct-ai-video-mid-scene-with-answers-in-under-a-second
