What AI Features Does Premiere Pro Have? A 26.5 Guide

In Summary Premiere Pro 26.5 (Adobe now calls the app Adobe Premiere) makes new media with two generative tools, Generative Extend and the Generative Media Tool. It transcribes and edits by text (Speech to Text, Text-Based Editing, Paper Edit), searches footage by description (media intelligence), cleans dialogue (Enhance Speech, Separate Crosstalk), fits music (Remix), masks subjects (Object Mask) and balances color (Auto Color). Generative Extend, the Generative Media Tool and Separate Crosstalk run in the cloud and spend generative credits. Generate Music, Generative Remove and the AI Assistant are in Premiere (beta) only. In the release version, nothing picks your soundbites for you. Paper Edit assembles the lines you select.

The short answer

Premiere Pro's AI features are generative tools, transcript tools, search tools and audio and picture helpers. Most of them work on media you already have. Three of them run in Adobe's cloud and charge generative credits: Generative Extend, the Generative Media Tool and Separate Crosstalk.

This guide covers Premiere on desktop 26.5.2, the current release (September 2026). Version 26.5 added the Generative Media Tool and Paper Edit. Adobe's help pages now call the app Adobe Premiere. They also document features that exist only in the separate Premiere (beta) app, so the table marks those rows (beta).

The table is compiled from Adobe's help pages. It lists the features those pages describe as AI, generative, Adobe Sensei or media intelligence. We found no Adobe page that lists every AI feature, so treat this as a map, not a complete inventory. "Added" comes from Adobe's What's new page. Where that page does not list the feature, the cell says "Not listed".

The features, with menus and limits

"None stated" means the Adobe page for that feature names no requirement. It does not promise the feature runs on every system.

Some features feel automatic but their Adobe pages do not call them AI, so they are not in the table: Auto Reframe (Sequence > Auto Reframe Sequence), Morph Cut (Effects > Video Transitions > Dissolve > Morph Cut), Scene Edit Detection (Clip > Scene Edit Detection), Auto Ducking (Essential Sound > Ducking > Generate Keyframes) and color matching (Comparison View, then Apply Match).

Generative tools: Generative Extend and the Generative Media Tool

Generative Extend fills the few frames you are missing: a reaction that ends a beat early, a transition that needs handle, room tone that runs out. Select the Generative Extend Tool in the Toolbar and drag the edge of the clip. Video clips should be at least 2 seconds long and audio clips at least 3. The new frames carry an AI-generated label. Right-click it for Regenerate or Revert to Original.

Adobe lists these limits on the Generative Extend overview:

• It cannot create or extend spoken dialog. Existing dialog is muted in the extension, and clips containing music are not eligible.
• One end only. It applies to the beginning or the end of a clip, not both.
• Lower spec than your source. Extensions above 30 fps are generated at 30 fps, in 8-bit and in SDR. The FAQ warns this can band smooth gradients like sunsets or skin tones.
• Mono and stereo audio only. Surround and 5.1 are not compatible.
• Known limitations with Speech to Text, Text-Based Editing, captions, media intelligence, markers, interchange formats (XML, AAF, EDL, OMF), multicam source sequences and Productions.
• Not available everywhere. Firefly is not offered in Russia, Belarus or China, and some K-12, education and enterprise accounts have limited or no access.

Version 26.5 accepts more source media: 360p to 4K UHD, any aspect ratio, and interlaced or non-square-pixel footage.

The Generative Media Tool is new in 26.5. Select it in the Tools panel, drag across the timeline, write a prompt in the generative task bar and pick a model. Video can come from Adobe Firefly or partner models such as Google Veo, Kling and Luma. Sound effects use Adobe's own model, and you can record your voice to guide timing and intensity. The task bar shows the estimated credits before you generate. Results land as editable clips, and the AI badge on a clip reopens its prompt.

Transcript tools: Speech to Text, Text-Based Editing, Paper Edit

Most of Premiere's assistive AI starts from a transcript. Open Window > Text, set the language and speaker labeling in the Transcript tab, and select Transcribe. English ships with Premiere. Other languages download from the Language menu, and Adobe lists 18 (supported languages).

From the transcript:

• Text-Based Editing trims the clips on the timeline as you cut and rearrange text.
• The filter icon in the Transcript panel finds filler words, pauses or one speaker, then Delete or Delete all. See remove filler words in Premiere Pro.
• Paper Edit (26.5) lets you select sentences or parts of sentences from anywhere in the transcript, preview them, then Create new sequence.
• Censored word list (Text panel > More options) bleeps or mutes listed words everywhere they appear.
• Translate Captions sends captions to Google Translate and Microsoft Translator and puts each language on its own track. See translate captions.

Paper Edit and Text-Based Editing make cutting faster, but you still choose every line. We compare that with an AI rough cut in Premiere Pro Text-Based Editing vs Eddie AI. For the first step, see transcribe video in Premiere Pro.

Search: media intelligence

Premiere analyzes footage in the background after import. Open the Search panel from the magnifying glass in the upper right and describe the shot. Results come back as Visuals, Audio, Text and Metadata. To find a look-alike shot, right-click a frame in the Program Monitor and choose Search Similar Frames.

Gotchas from Adobe's media intelligence FAQ and search guide:

• Visual search is English only.
• It does not identify people and does not read on-screen text.
• Transcript and metadata searches are exact text matches. Only visuals use natural-language descriptions.
• It looks at a smaller version of your footage as stills, so small details and fast motion are harder to find.
• Share the analysis. Before import, set Media Analysis & Transcription preferences > Cache analysis results for re-use to "Next to the media as a sidecar file". Other editors then skip the analysis.

Analysis and search run on your computer without the internet, and Adobe says it does not train on your footage or searches.

Sound and picture: Separate Crosstalk, Enhance Speech, Remix, Object Mask

Separate Crosstalk is for interviews and panels where people talk over each other. Trim the clip to the overlapping section, then choose Clip > Separate Crosstalk and Separate. Premiere makes a track for each detected speaker plus a background track, and disables the original. It is a premium feature: one generative credit per second, with a two-second handle added to each side that counts toward the cost. It works on one clip at a time. Adobe's cloud voice page says the audio is uploaded to Adobe, may involve voice embeddings, is not stored beyond the job and is not used for training. You confirm you have permission from the people in the recording before first use.

Enhance Speech lives in the Essential Sound panel. Select a dialogue clip, choose Enhance, then blend back some original sound with Mix Amount. See enhance speech in Premiere Pro.

Remix fits music to a length. Hold the Ripple Edit tool group, choose the Remix Tool and drag the clip edge. It keeps the start and end and rearranges the middle. See fit music to video.

Object Mask selects a subject with one click and tracks it, useful to blur a face. Its mask data sits in a folder next to the project file, named after the project with "Masks" added. Move that folder with the project, or the masks go offline.

What Premiere's AI will not do

• The release version will not pick your soundbites. Paper Edit assembles the lines you select. The AI Assistant, which can build stringouts and rough assemblies from a prompt, is in Premiere (beta) only, and Adobe does not recommend it for client work yet.
• It will not extend dialogue or music. Generative Extend mutes speech in the extension and refuses clips with music.
• It will not search every language. Visual search is English only, and Speech to Text covers 18 languages.
• Generative Extend output does not travel cleanly. Adobe lists XML, AAF, EDL and OMF interchange as known limitations.

Where Eddie fits next to Premiere

Eddie connects to Premiere through a dedicated extension, as the Premiere Pro workflow page shows. It reads clips, multicams and timelines from the Premiere bin, logs the footage and builds a rough cut. The sequence comes back into Premiere relinked to your original media.

That covers the gaps above:

• Choosing the lines. Eddie builds an edit from soundbites across one or more sources, with an optional target duration and pacing.
• Searching a project. Eddie searches transcripts for a phrase, topic or speaker and returns the segments with timestamps. It can also search the picture and sound for moments you describe, using credits per hour of footage searched.
• The handoff. Eddie exports a native Premiere Pro project or Premiere XML, and can check a Premiere export against the edit's timeline.

One thing to know: Eddie can clean wind, hum and noise from under the voice, but project files for editing applications keep the original sound. Run Enhance Speech in Premiere if you need clean dialogue in the timeline.

The step-by-step rough cut guide walks from bin to sequence. If you also cut in Final Cut, our Final Cut Pro AI features guide covers the same ground for version 12.4.

FAQ

Does Premiere Pro have AI features?

Yes. Premiere 26.5 includes Generative Extend, the Generative Media Tool, Separate Crosstalk, Speech to Text, Text-Based Editing, Paper Edit, Translate Captions, media intelligence search, Enhance Speech, Audio Category Tagging, Remix, Auto Color and Object Mask. Generate Music, Generative Remove and the AI Assistant are in Premiere (beta).

Which Premiere AI features use generative credits?

In the release version, Generative Extend (by resolution and frame rate), the Generative Media Tool (by model and settings) and Separate Crosstalk (one credit per second). In Premiere (beta), Generate Music and Generative Remove (40 credits per second) also use credits. Adobe says Speech to Text, Enhance Speech and Caption Translation are included with your membership.

Do Premiere's AI features work offline?

Some do. Media intelligence search runs on your computer without the internet. Generative Extend, the Generative Media Tool, Separate Crosstalk and caption translation need a connection. Speech to Text needs a language pack downloaded once for any language other than English.

Can Premiere Pro make a rough cut with AI?

In the release version, Paper Edit builds a sequence from the transcript lines you select. In Premiere (beta), the AI Assistant can create stringouts and rough assemblies from a prompt. Tools outside Premiere, such as Eddie, can propose the cut and send it back as a sequence.

Is Generative Remove available in Premiere Pro?

Only in Premiere (beta) at the time of writing. Adobe's page is titled "Generative Remove (beta)" and its steps start with opening your project in Premiere (beta).

Is my footage used to train Adobe's AI?

Adobe says no for Generative Extend, the Generative Media Tool, Separate Crosstalk and media intelligence. Its pages say your content is not used to train its AI models.

Sources

All checked November 6, 2026.

• Adobe: What's new in Adobe Premiere on desktop, release notes and What's new in Premiere (beta)
• Adobe: Generative Extend overview, Add frames using Generative Extend and Generative Extend FAQ
• Adobe: Generate media with the Generative Media Tool and Generative Media Tool FAQ
• Adobe: Separate overlapping speakers and About cloud-based AI voice features
• Adobe: Auto transcribe video using Speech to Text, Languages supported by Speech-to-Text and Download language packs from within Premiere
• Adobe: Overview of Text-Based Editing, Detect and delete pauses in transcripts and Create a sequence with Paper Edit
• Adobe: Translate captions and Clean audio with bulk mute or bleep
• Adobe: Media intelligence and Search panel, Search for media, Search for audio and Search for similar visuals
• Adobe: Apply Enhance Speech, Enhance Speech technical requirements and Automatically tag audio
• Adobe: Remix audio in Premiere and Remix tool considerations
• Adobe: Use Auto Color and Object masking in Premiere
• Adobe: Generate Music (beta) overview, Generative Remove (beta) and Premiere AI Assistant (beta) overview
• Adobe: Add Auto Reframe to sequences, Apply Morph Cut, Scene Edit Detection, Automatically duck audio and Match color between shots

---

Have hours of interviews and no assembly yet? Start a project for free, or read the Premiere Pro workflow first.