Changelog

New tools, improvements, and account updates: everything we ship, in one place.

  1. Improvement

    The talking-head cutter also cuts a false start inside a sentence

    When you stop in the middle of a sentence and start the same words again, the cutter now removes the abandoned words and keeps the clean take.

    Open the AI Video Editor
    • The repeated-take check now finds words you said, dropped and said again inside one sentence, for example "Today I want to I want to show you". It strikes through the first "Today I want to".
    • The cut appears in the transcript like any other retake. Click it to keep it, or switch off Cut repeated takes to keep every take.
    • The check cuts only a clear false start. A repeat made for emphasis stays, and a cut never leaves fewer than three words of the sentence.
  2. Improvement

    Long video to Shorts grades every clip on Hook, Flow, Value and Trend

    Each clip candidate in Long video to Shorts now shows four letter grades next to its score, so you see why a clip ranks where it does.

    Open the AI Video Editor
    • Every clip shows a grade for Hook, Flow, Value and Trend. Hover a grade to read what it measures.
    • The score is now the weighted total of the four grades: Hook counts most, then Value, Flow and Trend. Projects made before this release keep their score and show no grades.
    • Every clip has a thumbs-up and a thumbs-down. A favorite stays when you find more clips. A dislike removes the clip from the selection, and Find more clips never picks that moment again.
    • Press Command K (Control K on Windows) in a project, or press Search, to search the clip titles and everything said in the recording. Pick a result inside a clip and the clip scrolls into view and plays from that sentence. Pick a result outside every clip and a player opens at that moment.
    • When a recording has two or more voices, the clip options offer Color captions by speaker. The first speaker keeps the caption style color and every other speaker gets a color of their own, in the preview and in the rendered clip. Turn it off and the clip renders again without the colors.
  3. New Tool

    Video to video with Kling 4.0 Reference, and a Topaz video enhancer

    Two more models in the AI video generator start from your own footage: Kling 4.0 Reference builds a new clip from reference images and videos, and Topaz Video Upscaler sharpens or upscales a clip.

    Open the AI video generator
    • Kling 4.0 Reference takes up to 10 images and 5 videos of 2 to 30 seconds with a prompt, at 720p or 1080p for 5, 10 or 15 seconds. Name each file in the prompt as @image_1 or @video_1, or leave the names out and every file is referenced.
    • With a reference video, Kling 4.0 Reference bills a lower rate on twice the output length, or on the input plus the output when the videos are longer. The button shows the exact price.
    • Topaz Video Upscaler enhances a clip at 1x or upscales it 2x or 4x, with no prompt. It takes one MP4 or MOV clip of up to 60 seconds and 50 MB, billed per second.
    • New video to video and video enhancer pages compare the models, and the Kling 4.0 page explains the reference mode.
  4. New Tool

    AI motion transfer: Kling Motion Control and Wan 2.2 Animate

    The AI video generator now takes a video as input. Upload a character image and a motion clip, and Kling Motion Control or Wan 2.2 Animate makes the character perform the clip's moves.

    Open the AI video generator
    • Kling Motion Control runs Kling 3.0 or 2.6 at 720p or 1080p, follows the orientation of the clip (up to 30 s) or of the image (up to 10 s), and takes an optional prompt.
    • Wan 2.2 Animate has two modes: Move animates your image with the clip's motion at 720p, and Replace keeps the clip's scene and swaps its performer at 480p, 580p or 720p.
    • Clips can be MP4 or MOV. They upload straight to AgentReel's storage, and the limits match the provider's: 100 MB for Kling and Wan Move, 10 MB for Wan Replace.
    • These models bill by the second of the uploaded clip. The browser reads its length for the price on the button, and the server checks the stored file again before charging.
    • A new AI motion transfer page compares both models, and each has its own model page.
  5. Improvement

    AI Video Agent: size-budget export, speed ramps, video denoise, pixel diary

    The AI Video Agent can now compress a finished clip to an exact file size, apply speed ramps and beat-synced slow-mo, denoise a grainy clip from a measured noise level, and plan a pixel-character typewriter diary short from an episode script.

    Open the AI Video Agent
    • Fit size compresses a finished video to 8 MB, 10 MB, 16 MB WhatsApp, 20 MB Discord or WeChat, 25 MB email, or a custom byte count. It picks the largest ladder resolution that still has enough bits, two-pass H.264, tone-maps HDR to SDR, retries if the file overshoots, and reports size, resolution and SSIM. It is not a platform letterbox and not a GIF budget.
    • Speed ramp applies hero slow-mo, bullet, montage, flash-in, flash-out, jump or a raised-cosine dip, optionally landing ramps on beat times. Audio can be dropped, kept or replaced. It is not a beat flash, a multi-clip beat montage, a punch-in zoom or a last-frame hold.
    • Denoise video measures residual grain in flat regions, maps it to a denoise strength, and reports noise before and after. A clean clip gets a light pass. It is not a still enhance and not a document clean-up.
    • A pixel-character typewriter diary short (dialog box, pop-up cards, mouth flaps, 8-bit SFX) is an instruction pack for the code fallback. Character art is a still you provide or pixels drawn in code, never a paid image model.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  6. Improvement

    AI Video Agent: optical looks, window-pane tunnel intro, molecular films

    The AI Video Agent can now grade a clip as reeded glass, duotone, dither, bloom, VHS, CRT or radial colour fringing, open a clip through lighting perspective window panes, and plan a labelled molecular science film in Blender from a PDB structure.

    Open the AI Video Agent
    • Optical look is one skill with a style: fluted glass, duotone, ordered or error dither, bloom, VHS, CRT scanlines, or radial chromatic aberration. Audio is kept. It is not a linear RGB split, not a filmed-off-monitor push, not a risograph or CMYK halftone, and not a dissolve or wipe join.
    • Window-pane tunnel opening lights perspective panes, walks the camera forward, then sweeps a glowing edge to the same full-screen clip. Default first 9 seconds; shorter clips compress the motion. Original audio is kept. It is not a Ken Burns pan or a punch-in zoom.
    • A labelled molecular or protein science film from a brief plus a PDB or AlphaFold structure is an instruction pack for the code fallback, not a general Blender explainer. Public RCSB PDB only.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  7. Account

    Memory: see what AgentReel learned from your projects

    Your account now has a Memory page under My account. It shows the choices you usually make, such as frame, voice and caption style, learned from the projects you create and finish.

    • A value becomes a usual choice after at least 3 projects that mostly agree, and the page shows how many of your recent projects carried it.
    • You can set any choice yourself, learned or not, go back to automatic, write up to 20 short notes about your habits, switch memory off, or erase everything.
    • Memory is on for registered accounts and uses no credits. Guests are not remembered.
  8. Improvement

    AI Video Agent: blur-fill vertical that keeps the whole picture

    The AI Video Agent can now fill a 9:16, 1:1 or 16:9 frame with a blurred, slightly darkened copy of the clip and a sharp band of the whole picture, so gameplay and 16:9 captures stay fully visible instead of being cropped.

    Open the AI Video Agent
    • Blur-fill reframes a clip without cropping. The fill is a blurred pad; the original picture sits as a sharp band. Audio is kept.
    • It is not a talking-head crop, not black letterbox, and not a podcast cover audiogram.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  9. Improvement

    AI Video Agent: ASCII look, clean scans, still-image QC, Japanese study captions

    The AI Video Agent can now render a still or short clip as per-character-colour ASCII art, deskew photographed pages into clean frames, run a still-image quality check before a picture goes into a video, and burn Japanese study captions with kana and romaji.

    Open the AI Video Agent
    • ASCII look measures glyph ink from the sandbox font and can keep colour, edge characters, and a transparent background on stills. Animated input stays at its fps within a duration cap.
    • Clean scan deskews photographed pages, trims uniform margins, and writes PNGs plus a ZIP and a multi-page PDF, ready for a Ken Burns pan or a storyboard.
    • Still-image QC reports teal-orange split-tone, clipping, sharpness, noise and size as PASS or WARN. Video delivery QC stays on the existing check.
    • Japanese study captions stack kana and optional romaji above each line, with an optional translation. Inserting a narrated explainer slide at a named time is a recipe over existing cut, render, voice and caption skills.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  10. Improvement

    AI Video Agent: circular picture-in-picture, last-frame hold, RGB-split glitch

    The AI Video Agent can now put a circular talking-head over content, hold the last frame of a clip so a sentence can breathe, and give a clip an RGB-split glitch look.

    Open the AI Video Agent
    • Presenter-with-content now has a circular picture-in-picture layout, besides full screen, split screen and the rectangular inset.
    • A last-frame hold pads the end of a clip for a short beat and stretches silent audio to match. It is not a named-range cut and not a freeze inside the clip.
    • An RGB-split glitch shifts chroma channels and adds grain. It is not a risograph print and not a filmed-off-monitor look.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  11. New Tool

    Free watermark remover for photos and short clips

    Paint out watermarks, logos, timestamps and text overlays on a photo or a short clip you own, in the browser, with no upload and no credits.

    Open the watermark remover
    • The new Watermark Remover lives with the other free tools. Draw one or more boxes around a mark, or auto-detect a static corner logo, then download a cleaned PNG or MP4.
    • Images stay under 24 megapixels and 8192 pixels per side. Clips stay under 30 seconds and 1920 pixels on the long edge. Processing never leaves the tab.
    • A short notice next to the upload asks you to confirm you own the file or have the right to edit it. This is not a tool for taking marks off other people's work.
    • The AI Video Agent can run the same fill as the remove-watermark library skill in its sandbox (a purchase is required there, invite credits may pay).
  12. Improvement

    AI Video Agent: flash safety screen and racket-sport rally cuts

    The AI Video Agent can now screen a clip for photosensitive flashes before you publish, and split tennis, table-tennis, badminton or squash practice into one clip per rally.

    Open the AI Video Agent
    • A photosensitive flash screen (WCAG 2.3.1 three-flashes) reads the picture only and writes a PASS or FAIL report. It is an optional QC step after Code to Video, a beat cut, or any uploaded footage, not a compliance certificate.
    • Long racket-sport practice is split into one MP4 per rally by ball-impact sounds, dropping ball-picking and rest. This is not a silence cut, a music-beat cut, or a visual scene list.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  13. Improvement

    Code to Video: Chinese captions stay Chinese, and highlights wrap in reading order

    A Chinese Code to Video request, including a stick-figure explainer in the AI Video Agent, now writes captions and in-frame labels in Chinese. A highlighted word wraps inline instead of jumping to the end of the line.

    Open Code to Video
    • Ask in Chinese, or ask for a Chinese video, and every caption, title card and in-frame label is Simplified Chinese, even when the chat UI is English.
    • A highlighted keyword stays in the sentence in reading order. English wraps on spaces; Chinese wraps between characters.
    • Stick-figure explainers from the AI Video Agent follow the same language and caption rules.
  14. Improvement

    AI Video Agent: UI redline walkthroughs

    The AI Video Agent can now mark two UI screenshots with red ink and wipe from before to after, and it knows clay-look teaching shorts, narrated Ken Burns photo films, and cyan blueprint UX explainers.

    Open the AI Video Agent
    • Two screenshots of the same screen become a silent walkthrough: red boxes and notes on the before still, then a left-to-right curtain wipe to the after still. Missing screenshots are drawn first. This is not a Ken Burns pan of photographs and not a stripe-flip GIF.
    • A clay-look teaching short is Code to Video with a soft 3D clay look and a silent mascot in the brief, then an optional voice and captions. It does not call a video generation model.
    • A narrated cinematic over photos is the Ken Burns stills pass, then a voice, a mix, and captions. One continuous camera, not a slideshow of title cards.
    • A blueprint UX explainer names a cyan scan-line blueprint of the screen in a HyperFrames or Code to Video brief. Nothing is copied from a Figma starter.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  15. Seedance 2.5 draft mode

    Preview a Seedance 2.5 clip at 480p first, then upgrade the one you like to 1080p.

    Open the AI Video Generator
    • Turn on Draft mode under the Seedance 2.5 options. A 5-second draft costs 35 credits, the 480p price.
    • Press Upgrade to 1080p under the result within 7 days: 198, 395 or 593 credits for 5, 10 or 15 seconds.
    • The result card shows how many days are left to upgrade. A second press does not charge again.
  16. Long video to Shorts

    Turn one long recording into several vertical Shorts at once, in the AI Video Editor.

    Open Long video to Shorts
    • Upload one MP4 or MOV of up to 90 minutes. The editor transcribes it and ranks up to 8 clips of 15 to 60 seconds that start and end on a full sentence.
    • Each Short renders in 9:16, follows the speaker, and carries word-by-word captions and a hook title.
    • Adjust a clip by one sentence, edit its title, then download every Short in one ZIP file.
    • It needs purchased credits: about 15 credits to find the clips in a 10 minute video, then 2 credits a Short.
  17. Kling 4.0 Flash and FLUX 3 Klein

    Two fast, low-cost models join the AI video and image generators.

    Open the AI Video Generator
    • Kling 4.0 Flash makes 5, 10 or 15 second 720p clips with sound from text or a first frame, from 22 credits.
    • FLUX 3 Klein makes images at 1K, 2K or 4K and edits with up to 10 reference images, from 1 credit.
  18. Improvement

    AI Video Agent: silence cuts, green screen on stills, typed title cards, your own captions

    Four AI Video Agent editing steps now work the way they should, you can upload an .srt file, and audio plays right in the file list.

    Open the AI Video Agent
    • Cutting silence from a clip with sound no longer fails when more than one part is kept.
    • A green screen keyed over a still photograph now lasts as long as the take. It used to end after one frame.
    • Typed title cards work from a chat request, because the agent now writes the storyboard itself.
    • You can upload an .srt file. The agent can burn it as captions, and karaoke word captions take their timing from it.
    • GIF stickers show as pictures, and audio files play in the file list.
  19. Improvement

    AI Video Agent: lenticular flip cards

    The AI Video Agent can now turn two aligned stills into a looping lenticular or wiggle GIF, name easing curves in a HyperFrames or Code to Video brief, and punch-in on keyword times of an explainer.

    Open the AI Video Agent
    • Two same-pose stills can become a looping lenticular card: hold A, vertical stripe wipe to B, hold B, wipe back. A comic dynamic poster uses the same pair plus a title. One-still sticker actions stay on the existing GIF sticker pack.
    • Named easing curves such as ease-out-land, whip and duration-from-distance go in the Code to Video or HyperFrames brief. Nothing new is captured from a live meeting or an outside GIF factory.
    • A keyword-timed explainer clones or presets a voice, punch-in zooms on the keyword times, then burns word captions.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  20. Improvement

    AI Video Agent: Ken Burns stills, lookbook animatics, stick-figure explainers

    The AI Video Agent can now turn still photographs into a Ken Burns animatic, chain a character lookbook into that montage, make a stick-figure explainer, and wrap a product clip with kinetic captions, all without an AI video model.

    Open the AI Video Agent
    • One to six stills can be zoomed and panned into a silent 16:9, 9:16 or 1:1 clip. Consecutive stills take different pan paths so the montage does not loop one move.
    • A character lookbook is a canonical portrait, more views from that portrait as reference, then the Ken Burns pass. It does not call a video generation model.
    • A stick-figure explainer is Code to Video plus an optional voice and captions, not a whiteboard reveal of a painted still.
    • A kinetic product demo punches in on uploaded product footage and burns karaoke word captions. That is not a full product advertisement.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  21. Improvement

    AI Video Agent: GIF sticker packs and seeded grooves

    The AI Video Agent can now turn a character still into a looping GIF sticker pack, compose a short instrumental groove in a named style with a seed so catalog reels do not share one timeline, and name motion style packs in Code to Video or HyperFrames briefs.

    Open the AI Video Agent
    • A portrait or mascot still can become wave, heart, bow and idle looping GIFs, plus a zip of the pack. A mascot brand-short can follow from those stills.
    • A pop, lofi, cinematic, disco or chill groove can be synthesized locally. Each reel gets its own seed. Existing mix, ambient-pad and one-shot SFX skills stay the neighbors for ducking, noise and hits.
    • Motion style packs such as HUD, variety captions, Bauhaus, pixel and paper-shadow are named in the Code to Video or HyperFrames brief. Nothing new is captured from a shop or an outside gallery.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  22. Improvement

    AI Video Agent: stabilize, karaoke words, presenter layouts

    The AI Video Agent can now lock off shaky footage, list visual cuts, retune pitch without changing length, synthesize an ambient pad, master a platform frame, key a green screen, burn karaoke word captions, lay a presenter over content, and polish a screen recording, all in the same chat.

    Open the AI Video Agent
    • Handheld shake can be damped, then split on visual cuts and handed to the existing clip joiner.
    • A voice or stem can be raised or lowered by whole semitones without a speed change. A pink, brown, wind or drone pad can sit under a voice through the existing mix skill.
    • A finished clip can be letterboxed to TikTok, Shorts, Reels, square or YouTube and loudness-normalized as a last-mile master.
    • A green or blue backdrop can be keyed, then composed as full, split or picture-in-picture with a content clip. Word-timed karaoke captions highlight the active word.
    • An existing screen recording can be focus-zoomed at click timestamps with a synthesized click. The agent does not capture a browser.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  23. Improvement

    AI Video Agent: closer skill picks

    The AI Video Agent now steers the planner toward the tighter skill when two edits look alike, so a product ad, a B-roll cut, a voice clone, a privacy blur or a two-person crop land on the matching tool instead of a near-duplicate.

    Open the AI Video Agent
    • Product ads, voiceover-to-B-roll, voice clones, Code to Video, HyperFrames HTML, listen reports and privacy blurs have explicit pick rules.
    • Keyword matches are a hint, not a shortlist. The full skill list stays available.
    • Library skill descriptions now say when not to use a neighbor skill.
  24. Improvement

    AI Video Agent: speaker crops, mix gate, sketch and collage

    The AI Video Agent can now follow the person who is talking, check a mix against a delivery target, listen to footage, draw a still as a speed-painting, assemble a paper collage, burn bilingual captions, punch-in on a word, read on-screen cards, drop a sound effect and blur private regions, all in the same chat.

    Open the AI Video Agent
    • A two-person interview can become a 9:16 short that follows the talker, or a stacked pair of faces. A long clip can be scored for hooks and cut into several shorts with captions.
    • A mix can be measured (loudness, peak, ducking, a reference match) and graded SHIP, TWEAK or REMIX. The agent can then duck a music bed under the voice and land it on a social, YouTube, Spotify, podcast, EBU or ATSC target.
    • The agent can listen to a clip (pace, pauses, approximate speakers, a spectrogram), turn a still into a whiteboard reveal, or assemble generate-image cut-outs as stop-motion collage.
    • English-plus-Chinese captions keep WeChat and Douyin side margins. Punch-ins hit emphasis words. On-screen cards can be read and overlaid in another language. Short synthesized effects sit under the voice. Private boxes or faces can be blurred.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills. The connected-app catalog is unchanged.
  25. Improvement

    AI Video Agent: more in-chat edits

    The AI Video Agent can now reframe a clip (and follow a face when it sees one), burn styled subtitles, stack clips, cut silence, pulse to the beat, render a title-card storyboard, pull audio and draw a contact sheet, all in the same chat.

    Open the AI Video Agent
    • Ask for a vertical short and the agent can cut the talking head, reframe it to 9:16, 1:1 or 16:9 (a face when it sees one, otherwise the center), burn an .srt you already have and pull a thumbnail from a frame.
    • Burned subtitles can be an outline or a box, at the top, center or bottom. Chinese glyphs render when the sandbox has Noto CJK. A new language still uses the video translator.
    • Two clips can sit side by side or one over the other, or four in a grid. Dead air can be cut from a clip that is not a talking-head project. A music clip can flash on its beats.
    • A JSON storyboard of title cards becomes a video. A contact sheet and the audio track can be pulled from a reference so the agent can remake its structure on this site.
    • A cover of an existing video is still a frame of that video. A logo or title card is a PNG overlay. Animated titles still go through HTML render or Code to Video.
    • These edits run in the agent's sandbox and need a purchase (invite credits may pay), the same as other library skills.
  26. New Tool

    AgentReel MCP server

    The AI apps you already work in can now use AgentReel directly: add one address, sign in once, and ask for images and videos right there.

    Open the MCP page
    • The remote MCP server lives at https://agentreel.co/api/mcp and works with Claude Code, Claude, ChatGPT, Codex and Cursor. The new MCP server page and its setup guide show the steps for each app.
    • An app connects with an OAuth sign-in in your browser. AgentReel asks you to allow or deny every app that connects, and you never paste an API key.
    • A connected app gets 18 tools: AI images, AI videos, video templates, Code to Video, HTML render, video ads, B-roll cuts, the talking-head cutter, the AI video editor, the video translator, AI avatar video and voice clone, plus helpers for files, jobs, your balance and the help center.
    • Each tool spends the same credits as on its own page, and a failed job gives its credits back. Every app works in its own conversation of the AI Video Agent, where its files are saved.
    • The new Connected apps page in your dashboard lists every app you allowed and disconnects one at once.
    • Before your first purchase, a connected app can spend credits on Z-Image images, video templates and Code to Video only. The free B-roll trial and the free AI video stay on the website.
  27. Improvement

    Fuller model, tool, help and company pages

    The pages people read before they pick a tool now answer more of their questions on the page itself.

    Read about AgentReel
    • Model pages of a line with several versions, such as Nano Banana, Kling, Seedance, Wan and Grok Imagine, compare those versions in one table: resolutions, durations, reference images, aspect ratios and credits.
    • Resize Image, Image Converter, WebP to JPG, HEIC to JPG and Image Splitter list their sizes, formats and compatibility in tables.
    • Three blog comparisons now hold the four that repeated them: Nano Banana Pro against the original and 2.1 joins the Nano Banana 2 comparison, Veo 3.1 against Kling 3.0 Omni joins Kling 3.0, and Veo 3.1 against Seedance 2.0 joins Seedance 2.5. Old links open the right section.
    • The affiliate page explains how a referral is tracked, works through a yearly plan example and answers eight common questions.
    • About, Contact, the Refund Policy and the privacy guide say more: who makes AgentReel, a timeline of releases, every way to reach us, what happens to the credits of a failed task, and how your files are handled.
    • Each release in this changelog links to the page it changed.
  28. Improvement

    Help guides and model pages in fewer, fuller pages

    Eleven help guides became sections of the guide for the same tool family, and three model pages became sections of the newer model's page. Nothing was removed: every old link opens the right section, and every model is still selectable in the generators.

    • The YouTube clip maker, the talking-head cutter and Video Remix are sections of the AI Video Editor guide; the lyric video maker and Code to Video are in the video templates guide.
    • The video translator and AI avatar video are in the AI Video Generator guide, the TikTok carousel and the free image tools in the AI Image Generator guide, the desktop app in Getting started, and your works and billing in Credits and pricing.
    • Kling 3.0 is a section of the Kling 4.0 page, Seedance 2.0 of the Seedance 2.5 page and Nano Banana 2 Lite of the Nano Banana 2.1 page, each with its own price table, examples and questions.
    • The help center finder and the models catalog link straight to those sections, and links to the legal pages from a Chinese guide now open the right page.
  29. Improvement

    Guide pages folded into their tools

    Seven guide pages that repeated a tool now live as sections of that tool's page. Every tool works exactly as before, and every old link opens the right section.

    • The AI photo filter presets and the table of photo editing models are a section of the AI Image Generator; each preset still opens Nano Banana 2.1 with its prompt filled in.
    • Making an AI video from clips is part of the AI B-Roll Editor page; making a YouTube video from clips and combining clips on iPhone are sections of Combine Clips.
    • Clipping a video on iPhone and on a PC are sections of Clip a Video, with the same free in-browser clipper.
    • The lyric video template has one page, the Lyric Video Maker.
  30. Improvement

    Video templates grouped by kind

    The 33 video templates now sit on 8 pages, one per kind of video. Every template still renders exactly as before, and every old template link opens the page with that template already picked.

    Browse the video templates
    • Bar chart race holds every chart and data template: data story, comparison, animated counter, year in review, top 10 ranking, funnel animation and workflow diagram.
    • Kinetic type holds the text templates: quote card, countdown, split-flap board and typewriter text.
    • Product promo holds the launch and marketing templates: testimonial, event promo, what's new, pricing explainer, app preview, website launch and changelog video.
    • Logo intro has the YouTube outro, Chat story has the notification pop and the AI chat reveal, Slideshow has before and after, collage explainer and travel map route, and Lower third has breaking news.
    • Pick another template in the picker and the address follows, so a shared link opens the same template.
  31. Improvement

    Related free tools share one page

    Free image tools that answer the same need now live on one page, one tab per tool. All 25 tools still work exactly as before, and every old link opens the right tab.

    Browse the free tools
    • Sharpen Image has an Unblur Image tab, and Photo Enhancer has a Clear Image tab.
    • Background Remover has Product White Background and Signature Background Remover tabs.
    • JPG to WebP also converts PNG and JPEG to WebP; WebP to JPG also converts WebP to PNG and JPEG; Image Converter has PNG to JPG and JPG to PNG tabs.
    • Resize Image has an Image Cropper tab with all five crop and split modes.
    • The address keeps the tab you pick, so a shared link opens the same tool.
  32. New Tool

    AI photo filter

    A new AI photo filter page puts twelve one-click looks on top of the AI image generator. Pick a look, upload a photo, and an edit model re-lights and re-grades it while the face, pose and framing stay the same.

    Open the AI photo filter
    • Twelve presets: golden hour, blue hour, CCD camera, old phone camera, film halation, rim light, butterfly lighting, pink sunset, warm film, lo-fi dusk, cobalt blue and water reflection. Each opens Nano Banana 2.1 with the prompt filled in.
    • A table lists every model that can edit an uploaded photo, with its reference limit and the credits for one photo, read from the live price list.
    • Filtering your own photo uses purchased credits; free credits still pay for Z-Image only.
    • The page is linked from the AI image generator menu, its guide and the photo enhancer.
  33. Account

    AgentReel affiliate program

    Creators can now earn 30% recurring commission for 12 months by recommending AgentReel. The full terms are on the new Affiliates page.

    See the affiliate program
    • 30% of what each referral actually pays for 12 months, on subscriptions, annual plans and credit packs, with a 60-day cookie.
    • Monthly NET-30 payouts by PayPal or Wise from $50, tracked by Tolt.
    • The /affiliates page lists the rules, an earnings example and who the program is for, and is linked from the footer.
  34. Improvement

    New models: Kling 4.0, Nano Banana 2 Lite and Vidu Q4 in 4K

    Kuaishou's Kling 4.0 joins the AI video generator, Google's Nano Banana 2 Lite joins the AI image generator, and Vidu Q4 now renders up to 4K and takes up to 15 reference images. Each has its own model page.

    Open the AI video generator
    • Kling 4.0 makes clips with native sound from a prompt, or from a first and a last frame, at 720p or 1080p and 5, 10 or 15 seconds.
    • Nano Banana 2 Lite generates or edits one 1K image for 1 credit, with up to 10 reference images and 15 aspect ratios.
    • Vidu Q4 adds 2K and 4K, and two or more images now switch it to reference-to-video with a choice of 16:9, 9:16, 1:1, 4:3 or 3:4.
    • New model pages for Kling 4.0, Nano Banana 2 Lite and Vidu Q4, linked from the generators, the model library and the Kling 3.0 and Nano Banana pages.
  35. Improvement

    New model page: GPT Image 2.5

    OpenAI's GPT Image 2.5 has its own page, with settings, the full credit price table, a comparison and real example images.

    Open the AI image generator
    • The GPT Image 2.5 Sunburst page explains when to pick Sunburst or Flare, with every frame, resolution and credit price read from the image generator's catalog.
    • Image model pages now show real example images with the prompts that made them when the showcase has them for that model.
    • The AI image generator, TikTok carousel and Nano Banana Pro pages link to the new guide.
  36. New Tool

    AgentReel for iPhone: the native iOS app is in testing

    AgentReel for iPhone is a native iOS app with every AgentReel tool, built for the phone instead of a web page in a frame. It is now in testing ahead of its App Store release.

    • Every tool runs in the app: the AI image and video generators, video templates, Code to Video, TikTok carousels, the AI video agent, the B-roll cut, the talking-head cutter, the clip maker, the AI video editor, video remix, the video translator, AI avatar videos and the AI video ad generator.
    • Sign in with Apple, with an email code or as a guest. A guest who signs in keeps everything it made.
    • Credits are bought through the App Store, in the same packs and plans as on the site, and one balance serves the web and the app.
    • Free credits work as on the web: they pay for Z-Image pictures, video templates and Code to Video renders.
    • The 25 free image tools run on the phone itself, and finished work saves to Photos or posts to TikTok and YouTube.
    • Share links to agentreel.co open the work in the app.
  37. Improvement

    New model pages: Gemini Omni 1.1 Flash, Grok Imagine 1.5 and Nano Banana

    Four more models have their own page with settings, the full credit price table and a comparison: two video models and the first two image models.

    Open the AI video generator
    • Gemini Omni 1.1 Flash and Grok Imagine Video 1.5 join the video model pages, each with its frames, resolutions, lengths and credit prices from the generator's catalog.
    • Nano Banana Pro and Nano Banana 2.1 are the first image model pages; their Generate button opens the AI image generator with the model selected.
    • The Gemini Omni 1.1 Flash page plays a real example clip with the prompt that made it.
    • The model overview guide now covers the new video models, and the AI video and image generator pages link to the new guides.
  38. Improvement

    AI model pages: Veo 3.1, Kling 3.0, Seedance, Wan 3.0 and MiniMax H3

    Six AI video models now have a page with its settings, its full credit price table and a comparison with the other models, plus a model overview page.

    Open the AI video generator
    • Model pages for Veo 3.1, Kling 3.0, Seedance 2.5, Seedance 2.0, Wan 3.0 and MiniMax H3 list the aspect ratios, resolutions, durations and prompt length the generator accepts.
    • Each page shows the credit price of every setting combination, read from the same catalog the generator charges from.
    • The Generate button on a model page opens the AI video generator with that model already selected.
    • The model overview lists every video and image model with its developer and credit range.
    • The Seedance 2.0, Wan 3.0 and MiniMax H3 pages play example clips made on AgentReel, each with the prompt that made it.
    • Every model page and the overview show the day they were last updated, and the overview adds a short guide to choosing a video model by format, control, sound, length and budget.
  39. Improvement

    Footage sources in the AI B-Roll Editor, the desktop app and the AI Video Agent

    Bind clips to a stretch of your script, and only those clips appear there. The AI Video Agent now cuts B-roll too, from your uploads and from the videos it makes.

    • In the AI B-Roll Editor and in AgentReel for Mac and Windows, add a footage source: pick the phrases, then the clips. Changing the sources matches the footage again.
    • Ask the agent for a B-roll video and say which clips go with which paragraph of your script. Director notes in (parentheses) guide the picture and are never read aloud.
    • A video the agent makes in one step can be a clip of a B-roll cut in a later step.
    • A B-roll cut in the agent is priced and charged as on the AI B-Roll Editor page and needs a purchase.
  40. New Tool

    AI Video Agent: a code tool for tasks no skill covers

    When none of its tools fits a video, image or audio task, the agent hands the task to an AI coding assistant inside a sandbox and saves a method that works as a skill for everyone.

    Open the AI Video Agent
    • The code tool needs a purchase; invite credits can also pay for it. The estimate is 200 credits and at most 600 are held while it runs; a run keeps only the credits for the work it did, even when it fails, and gives back the rest.
    • One person can start up to 10 code runs a day.
    • When the code tool calls the image or video generator, each image or clip is charged at that generator's own price.
    • A saved skill is ready from the next conversation on, for you and for every other user.
    • The terms of service cover the files the code tool makes and the skills it saves: /terms-of-service#agent-code.
  41. New Tool

    AI Video Agent: describe the video, the agent plans and makes it

    Describe the video you want on the homepage or at /agent, add your own clips if you have them, and the agent plans the steps, runs the tools and shows the finished video with a download.

    Open the AI Video Agent
    • The agent runs the image, video, template, Code to Video, voice, translator, avatar, video ad, talking-head and editing tools, and checks the credit estimate against your balance before it starts.
    • Every step appears as a card with its status, its estimate and what it cost.
    • An estimate above your balance pauses with a purchase button; nothing is charged.
    • Free credits pay for Z-Image, video templates and Code to Video in the agent too; other models and tools need a purchase.
    • Talking to the agent costs no credits: 5 requests a day before your first purchase, 50 after.
    • Conversations are private and listed on /agent.
  42. New Tool

    AI Video Editor: edit your own video by typing what to change

    Upload a video and say what to change in plain words. AI cuts the pauses, adds captions, reframes to vertical or square, grades the look and adds titles, music or a logo, then lets you ask for the next change.

    Open the AI Video Editor
    • Open AI Video Editor from the AI Video menu.
    • Upload a video up to 10 minutes long, plus music, a voice track or a logo if you want them.
    • Every edit shows four preview stills first; ask for changes again and again, or go back to an earlier version.
    • Editing and previews are free. Rendering the MP4 uses purchased credits, from 15 credits for up to 30 seconds.
  43. New Tool

    AI Avatar Video: a talking presenter from one photo

    Upload one portrait and a script or a recording. AgentReel animates the face so it speaks your words, with optional burned captions.

    Open AI Avatar Video
    • Open AI Avatar Video from the AI Video menu.
    • Type a script for an AI voiceover, or upload your own MP3, WAV or M4A recording.
    • Choose 720p or 1080p; the price follows the exact length of the speech and uses purchased credits.
    • Download the MP4 plus SRT and VTT captions. Any failure before delivery refunds the credits.
  44. New Tool

    Video Translator: subtitles and dubbing in 16 languages

    Translate a video of up to 10 minutes into 16 languages as SRT and VTT subtitles, burn them into the picture, or dub it with AI voices that keep the speaker's voice.

    Open the Video Translator
    • Open Video Translator from the AI Video menu.
    • Subtitles and dubbing use purchased credits, priced by the length of the video.
    • Any failed translation refunds its credits.
  45. New Tool

    Video Remix: reuse what makes a short video work

    Upload a short video you like. AgentReel breaks down its hook, beats, shots, pacing, captions, sound and retention tricks, then writes a new script and storyboard for your topic.

    Open Video Remix
    • Open Video Remix from the AI Video menu.
    • A breakdown and a remix cost 13 purchased credits each, refunded if a step fails.
    • Send the new script to Code to Video, the AI Video Ad Generator or B-roll Cut in one click.
  46. New Tool

    Talking-Head Cutter

    Upload a talking-to-camera recording; pauses, fillers and retakes are cut automatically, and you can restore any cut before rendering.

    Open the Talking-Head Cutter
    • Transcript editor with every cut struck through
    • Jump-cut render with optional captions and punch-in zoom
    • Editable timeline export for Premiere, Resolve and Final Cut
    • Analysis and rendering use purchased credits; a failed step refunds them.
  47. New Tool

    Code to Video: a video for anything no template covers

    Describe the video you want. AI writes it as code, checks four preview frames, fixes what looks wrong and renders an MP4.

    Open Code to Video
    • Open Code to Video from the AI Video menu or the Templates menu.
    • Writing the code and the preview frames is free. Ask for changes in plain words as often as you like within the daily limit.
    • A render costs 15 credits up to 30 seconds, and free check-in credits can pay for it.
    • Every template render now costs at least 15 credits, the same as one daily check-in.
  48. Improvement

    AgentReel Docs: a guide for every tool

    A new help center explains what every tool makes, its limits and what it costs, in English and Chinese.

    Read the help guides
    • Type what you want to make in the "Can AgentReel do this?" search to find the right tool.
    • Every guide lists file limits, credit prices and the free options.
    • Open Docs from the footer of any page.
  49. Improvement

    Download from a shared video page

    A public share page now has a clear download button, and the share dialog copies the page link.

    • Viewers of a shared video can save it in one click.
    • Copy the public page link straight from the share dialog.
  50. New Tool

    AgentReel for Mac and Windows

    A desktop app for the AI B-Roll Editor that works with the footage on your own computer.

    Read the desktop app guide
    • Cut long footage without uploading every file first.
    • Sign in with your AgentReel account. Credits work the same as on the web.
    • Download it from the site header or footer.
  51. Account

    Invite a friend, you both get 100 credits

    Share your invite link. When a friend signs up with it, you both get 100 credits.

    Invite a friend
    • Find your link on the credit history page, or under Invite friends in the menu.
    • Invite credits pay for any model and almost every tool.
    • You earn credits for up to 5 friends.
  52. Improvement

    AI video ads: your own voice, silent shots and on-screen text

    The AI Video Ad Generator now reads your ad in your own cloned voice and follows timed shot lists more closely.

    Open the AI Video Ad Generator
    • Pick one of your own cloned voices next to the six preset voices.
    • A shot can run without voiceover, for a silent hook.
    • Give each shot its own on-screen text.
    • Footage you already shot can go to the clip maker for trimming.
  53. New Tool

    New tool: YouTube clip maker

    Share just one part of a YouTube video, or trim your own video file right in the browser.

    Open the YouTube Clip Maker
    • Set a start and an end on a YouTube link to get a timestamped share link and an embed code.
    • Trim a video file to MP4 in the browser. It is free and nothing is uploaded.
    • Add an .srt or .vtt file to burn the subtitles into the clip.
    • Describe a moment, and AI highlight search finds up to five matching ranges for 15 credits.
  54. Improvement

    B-Roll Cut is now AI B-Roll Editor

    Give it messy, unsorted clips and a script, and get back one finished video with voice, captions and music.

    Open the AI B-Roll Editor
    • The same tool, with a name that says what it does.
    • It now sits first in the site header.
  55. Account

    Live chat from any page

    The help button in the corner now opens live chat or the feedback form.

    • Ask a question and get an answer right away.
    • A red dot shows when a reply is waiting.
  56. Account

    Your first B-roll video is free

    An account that has never purchased can cut one B-roll video at no cost.

    Open the AI B-Roll Editor
    • Up to 60 seconds and 50 clips.
    • One AI voiceover is included, or use your own recording.
    • After the trial, credits pay for every next video.
  57. Improvement

    Close the tab while your B-roll video is cut

    Your cut keeps running on our side after you leave, and we email you when it is ready.

    Open the AI B-Roll Editor
    • Start the cut, close the page and come back later.
    • The email links straight to the editor to watch and download.
    • If a cut stops, the email says why and refunds the failed step.
  58. Improvement

    B-roll videos with your own voice

    Upload a recording of yourself, or clone your voice once and let it read any script.

    Open the AI B-Roll Editor
    • Upload an MP3, M4A or WAV voiceover up to 50 MB. Captions follow your words.
    • Record or upload 30 seconds to 3 minutes of your voice to create your own AI voice (paid accounts).
    • Every new voice saves a version, and you can switch back to an earlier one at no cost.
  59. Improvement

    B-roll uploads keep going while you browse

    Large footage uploads continue when you move to another page, and pick up again after a reload.

    Open the AI B-Roll Editor
    • A small progress panel follows you across the site and links back to your project.
    • After a reload or a closed tab, the editor offers to resume the upload.
    • Files that already finished are skipped when you pick the same batch again.
  60. Account

    Prices in six more currencies

    The pricing page now shows prices in Brazilian real, Mexican peso, Indian rupee, Polish zloty, Thai baht and Korean won.

    See the prices
    • Pay in BRL, MXN, INR, PLN, THB or KRW.
    • The plans and credits stay the same in every currency.
  61. Improvement

    Six new caption styles in the B-roll editor

    Pick the caption look that fits your channel: thick outline, shadow, yellow, label or rounded.

    Open the AI B-Roll Editor
    • Thick outline and shadow styles stay readable over bright footage.
    • Yellow, label and rounded styles match the look of popular short-form channels.
    • Switch the style at any time and render again.
  62. Improvement

    Lyric video maker page

    The lyric video workspace now has its own page at /ai-video-templates/lyric-video.

    Open the lyric video template
    • Open /ai-video-templates/lyric-video to start a lyric video from your own song in one step.
    • The page explains beat analysis, lyric alignment, styles and pricing, with answers to common questions.
  63. Improvement

    AI video ads now follow your whole brief

    Paste your full brief or script into the AI Video Ad Generator, and AI fills the form for you and keeps your words.

    Open the AI Video Ad Generator
    • A new box at the top reads your whole brief and fills every field. You can edit any field before you start.
    • Your own voiceover script is spoken exactly as written, and your visuals, style and structure come first.
    • The fixed copy rules are gone: first person, numbers and your product names are all fine.
    • The video runs as long as your script, up to 120 seconds. Scenes without a recurring person skip the character step and its cost.
    • The voiceover is now recorded before any image, so the real length is checked before you pay for pictures. A shot longer than one clip is split at its natural pauses instead of failing.
    • Choose the frame: vertical 9:16 (with a square crop), square 1:1 or horizontal 16:9 for YouTube and presentations, at the same price. The brief reader picks it for you when your brief names where the video plays.
  64. New Tool

    New tool: AI B-roll cut matches your clips to a voiceover script

    Paste a script, upload your footage, and AI picks a short shot for every phrase, then renders one MP4 with captions.

    Open the AI B-Roll Editor
    • Upload up to 500 .mp4 or .mov clips and paste a script in English or Chinese, up to about 8 minutes of speech.
    • AI reads every clip, then picks footage for each phrase. Shots of under a second are cut on the beat of the captions.
    • Check the timeline phrase by phrase, then render a 16:9 or 9:16 MP4 with burned-in captions. The video has a silent track for now.
    • B-roll cuts need purchased credits. The editor shows the price before it starts, and a failed step is refunded.
  65. New Tool

    Lyric video maker for your own song

    Upload a song you have the rights to, paste the lyrics, and get a lyric video that moves on the beat.

    Open the lyric video template
    • Pick the Lyric video template, upload your song and confirm you own the rights or have permission.
    • Analyze beats for free. The tempo is detected and you can correct it by hand.
    • Align the lyrics for 5 credits so every word lights up as it is sung. Without alignment, lines are spread evenly.
    • Choose a 10 to 120 second clip and a bold, neon or minimal style. Backgrounds cut on the beat.
  66. New Tool

    Collage explainer template

    A paper-cutout explainer: each image arrives as a hand-cut piece with tape, a headline and a sticky note.

    Open the collage explainer template
    • Pick Collage explainer and add 3 to 8 images, one for each point you want to make.
    • Choose kraft, white or black paper. The video ends on a collage of every cutout.
    • Add narration to read one line per image, with captions styled as a paper label.
  67. New Tool

    Website to video: a launch video from any public URL

    Paste a website link and we capture the page, write the copy and build a launch video.

    Open the website launch template
    • Open the Website launch template and paste a public URL into the capture box.
    • We take desktop and mobile screenshots and read the page text, free of charge.
    • Use the desktop shots for a scrolling launch video, or the mobile shots for the App preview template.
    • Pages behind a login cannot be captured. Upload your own screenshots instead.
  68. New Tool

    Changelog video from a GitHub pull request or release

    Paste a public GitHub pull request or release link and get a ready-to-render changelog video.

    Open the changelog video template
    • Open the Changelog video template and paste a GitHub link into the import box.
    • We read the title, the changes and the stats, then write the scenes for you.
    • Each change gets a tag such as feature, fix or performance, with counters and an end card.
  69. Improvement

    Template videos can now talk: narration, music and captions

    Add an AI voiceover, a music bed and word-by-word captions to any AI video template.

    Browse the video templates
    • AI writes a narration that fits the length of your video. Edit the lines, then pick a voice.
    • Describe a mood and AI makes a music bed. It ducks under the voice and the mix is set to streaming loudness.
    • Captions light up word by word, at the bottom or the top, or turn them off.
    • Narration and music need purchased credits. The button shows the price before you generate.
  70. New Tool

    New template: map route animation

    Turn a trip into a short travel map video. Name the stops, and AI draws the route across a world map.

    Open the map route template
    • List 2 to 8 places in travel order. AI fills in the coordinates, and you can correct them.
    • Pick a plane, car, train or boat. The line draws one leg at a time and the marker follows it.
    • Choose a dark, light or paper map, in 9:16, 16:9 or 1:1.
    • The map is drawn from public-domain Natural Earth data, so the video is yours to post anywhere.
  71. Improvement

    AI video ads from 15 to 60 seconds, from about 220 credits

    Choose how long your ad runs and how many versions of every shot AI makes. Short ads now cost much less.

    Open the AI Video Ad Generator
    • Pick 15, 30, 45 or 60 seconds. The script, the voiceover and the number of shots follow the length you chose.
    • Pick 1, 2 or 3 versions of every still and clip. One version is the cheapest; more versions give you more to choose from.
    • A 15-second ad with one version costs about 220 credits, so Pro covers up to 4 short ads a month and Ultra up to 9. The form shows the estimate of every choice before you start.
    • Long shots now use 8-second clips, so 45 and 60 second ads cost less than with 10-second clips.
  72. Improvement

    28 AI video templates, each with its own page, in three frames

    The AI video templates tool grows from 6 to 28 templates, and every template now renders in 9:16, 16:9 or 1:1.

    Browse the video templates
    • 22 new templates: bar chart race, lower third, breaking news, notification pop, AI chat reveal, YouTube outro, before and after, quote card, stat counter, year in review, top ranking, funnel, workflow diagram, split flap, typewriter, testimonial, event promo, countdown, feature update, pricing explainer, app preview and comparison.
    • Pick a frame for any template: vertical 9:16 for TikTok, Reels and Shorts, wide 16:9 for YouTube and slides, or square 1:1 for feeds. The six original templates support all three too.
    • Every template has its own page at /ai-video-templates/<name> with the editor already set to that template, a guide and an FAQ.
    • The Video Templates link in the header opens a menu of all templates by category: marketing, data, social, broadcast and text.
    • Templates that show photos take your own uploads or your AI images, JPG, PNG or WebP up to 10 MB.
  73. Improvement

    Your real product in your video ad

    Upload one to three photos of a physical product and the AI Video Ad Generator shows that product, not an invented one, in the shots where it appears.

    Open the AI Video Ad Generator
    • Brand kit: a new Product photos slot takes up to three photos (PNG, JPG or WebP, 10 MB each). Add them before the script is written.
    • The script marks the one or two shots that show the product (the turn and the payoff), and the storyboard shows a Product badge on them.
    • Those stills are generated with your photos as references, so shape, colors and materials match. Redo and fix keep the references, and the clips start from the stills.
    • Prices do not change: a still costs the same with or without references.
  74. Improvement

    Review the voiceover and clips of your video ad, and change a finished ad

    In guided mode the AI Video Ad Generator now also stops at the voiceover and at the clips, and every finished ad can go back to the clips for another export.

    Open the AI Video Ad Generator
    • Voiceover: listen to three takes with their length and tempo, see whether the sigh is audible, pick one, or record another take.
    • Clips: pick one of two clips per shot or keep the still, then set where each clip starts and how fast it plays. A badge shows whether the clip fits its slot, checked by the same rule as the final render.
    • Choose the music bed, switch sound effects on or off, change where captions break or hide a caption, and mark which parts of your screen recording fill the app shot.
    • Build a rough cut to watch your picks together before the final render. Change and export again goes back to the clips; each new export costs the price of one final render.
  75. Improvement

    Review the character and storyboard of your video ad

    In guided mode the AI Video Ad Generator now stops at the character and at the storyboard, so you pick every image before any shot is animated.

    Open the AI Video Ad Generator
    • Character: pick one of two versions, make another, or fix one thing in the picked image, such as the clothing.
    • Storyboard: one column per shot with two images each. Pick an image, lock the shots you like, and redo a shot with your own prompt or one targeted fix.
    • Every image shows which part the square 1:1 video keeps, so faces and the key action stay in frame. A new or fixed image costs the price of one image, and failed images are refunded.
  76. New Tool

    AI Video Ad Generator makes a 30-second ad for your product

    Describe your product, approve one of three AI concepts and its script, and get a 27 to 31 second video ad with voiceover, animated shots, music and captions in 9:16 and 1:1.

    Open the AI Video Ad Generator
    • Automatic mode: after you approve the concept and script, AI draws the character and shots, records the English voiceover, animates each shot, adds music and sound effects, and renders both formats for your final review.
    • Brand kit: your logo, colors, a built-in or licensed font, an end card, a screen recording of your app, a character photo with consent, and your own music track.
    • Concepts and scripts are free. The rest is charged step by step from purchased credits, the editor shows the estimate first, failed steps are refunded, and you confirm before an ad passes 1.5 times its estimate.
  77. New Tool

    AI Video Templates turn one sentence into a vertical MP4

    Pick one of six templates, describe your video, and AI writes the text, colors and timing. Check the frames, edit anything, then render a 1080x1920 MP4.

    Browse the video templates
    • Templates: product promo, kinetic typography, logo intro, data story chart, chat message story and image slideshow.
    • Writing with AI and the frame check are free; AI looks at four still frames and fixes text that is cut off or hard to read before you pay.
    • Rendering costs 10 credits up to 20 seconds, plus 4 credits per started 10 seconds; a failed render is refunded automatically.
  78. Improvement

    Download saves generated images and videos directly

    The Download button on the AI image and video generators now saves the file to your device instead of opening it in a new tab.

    • Works the same on desktop and on mobile browsers, including Safari on iPhone.
    • Files are named agentreel-<id> with their original format, PNG, JPG, WebP or MP4.
  79. New Tool

    22 more free image tools, grouped by category

    The free tools page now holds 25 browser tools in four categories: enhance, background, convert and crop.

    Browse the free tools
    • Enhance and upscale: Photo Enhancer with six subject profiles, Anime Upscaler, Image to 4K (2K, 4K or 8K), Denoise, Unblur and Clear Image.
    • Background: a white background for product photos sized for online stores, and a transparent PNG from a handwritten signature.
    • Convert and compress: an image converter with quality control and a ZIP download, eight one-click format pairs between PNG, JPG and WebP, HEIC to JPG, Resize Image and Compress Image.
    • Crop and split: crop to exact ratios or a circle, and split long screenshots, grids or fixed tiles into a ZIP.
    • The free tools page and the sidebar now group tools by category. Like every free tool, they run in your browser: no sign up, no upload and no watermark.
  80. New Tool

    AI Image Generator with 20 models

    Create and edit images with 20 leading AI models, all in one workspace.

    Open the AI Image Generator
    • Pick a model, write a prompt, and choose the aspect ratio and quality; the Generate button shows the credit price before you start.
    • Add up to the model's limit of reference images by upload, drag and drop or paste to edit or restyle a picture.
    • Start without an account: guests wait about 15 seconds in a free queue before each image, and buying credits skips the wait.
    • Credits for a failed image are refunded automatically, and your latest images stay in the history strip.
    • Tool pages, including the free image tools, now share one workspace with a sidebar for quick switching.
  81. New Tool

    Free image tools: sharpen, upscale and remove backgrounds

    Three free image tools now run right in your browser, with no sign up, no watermark and no upload.

    Browse the free tools
    • Sharpen Image: Light, Medium or Strong sharpening for JPG, PNG and WebP.
    • Image Upscaler: enlarge photos 2x or 4x.
    • Background Remover: download a transparent PNG or WebP.
    • Find all three under Free Tools in the header, footer and homepage.
  82. Account

    Daily check-in: 15 free credits

    Sign in and claim 15 credits from the daily check-in banner. The credits stay in your balance until you use them.

    • One claim per account per UTC day; resets at 00:00 UTC.
    • Check-in credits do not expire and are separate from the seven-day free allowance.
  83. Improvement

    Product updates in one place

    The changelog now includes the latest releases, with a permanent link in the footer.

    • Read dated release notes for new features and improvements.
    • Release notes are available in English and Chinese.
  84. New Tool

    AgentReel is live

    AgentReel launches as a home for AI-made videos, slides and carousels, working right in your browser.

    • Turn an idea into a short video ready for social media
    • Build slides and carousels from text and images
    • Download and share your results
  85. Improvement

    Your dashboard and credit history

    See your balance, orders and every credit change in one dashboard.

    • Track subscription and pack credits separately
    • Review past orders and invoices
    • Manage your account settings in one place
  86. Account

    20 free welcome credits and flexible plans

    Every new account gets 20 free credits at signup to try the tools.

    • 20 free credits at signup, no card required
    • Subscription plans with monthly credits and rollover
    • One-time credit packs that never expire

Ready to try the latest tools?

Explore AgentReel