OpenClipsAI (@openclipsai) header
OpenClipsAI (@openclipsai) profile picture

Every top AI video model in one studio, from Seedance 2.5 to Kling and Veo. Built for creators, brands and teams 🎬 Free credits when you refer a friend.

Joined July 2026
Photo by OpenClipsAI (@openclipsai)Photo by OpenClipsAI (@openclipsai)Photo by OpenClipsAI (@openclipsai)Photo by OpenClipsAI (@openclipsai)Photo by OpenClipsAI (@openclipsai)Photo by OpenClipsAI (@openclipsai)Photo by OpenClipsAI (@openclipsai)Photo by OpenClipsAI (@openclipsai)Photo by OpenClipsAI (@openclipsai)Photo by OpenClipsAI (@openclipsai)
We tried the same prompt on 4 AI models: build a storm at sea from scratch, in code 🌊 GPT-6.2 Astra, GPT-6.2 Sol, Claude Opus 5.5 and Claude Fable 5.2. Which boat makes it home? Try them all on OpenClips.
1
42
We tried the same prompt on 4 image models: a 1950s diner on Mars 🚀 Seedream 5.5, GPT Image 2.6, Nano Banana Pro and Grok Imagine 2.0. Which one would you grab a coffee in? Try them all on OpenClips.
57
Ideogram 4.6 is out now ✂️ One photo. 7 edits.A dog shows up. Goggles go down. The coat turns black. The neon says OPEN LATE. He gets a fade. It starts to rain. And everything outside each edit stays exactly where it was. 0 pixels moved, we measured it. Ideogram 4.6 is coming soon to OpenClips. The demo runs on Ideogram 4.5 Precise Edit, which already works on OpenClips today, and inside Claude or ChatGPT through our MCP. Sign up free and the first 30 credits are on us to get you started 💈
1
60
Claude can now edit your videos. Type "make this clip golden hour", "upscale it to 4K" or "dub it into Spanish" and OpenClips does it right in the chat. Connect it in 4 steps: Settings > Connectors > Add custom connector > mcp.openclips.ai. 30 free credits to start.
1
47
We tried one prompt on 4 image models: Seedream 5.0 Pro, Nano Banana 2.1, FLUX 3 and GPT Image 2.5. Look at the hands, the tiny gears and the sign on the window. Only one got the text right. Which one would you pick? Try them all on OpenClips
2
7
259
Prompt 👇 Cinematic photo of an old watchmaker's workshop on a snowy winter night. An elderly woman with silver hair and a jeweler's loupe over one eye repairs an open pocket watch with fine tweezers, her hands sharp and in focus. Hundreds of tiny brass gears, springs and screws are laid out in neat rows on a green felt mat. A warm brass desk lamp is the only light, steam rises from a cup of tea beside her. Behind her, a frosted shop window with snow falling outside, and the hand-painted gold letters on the glass read "REPAIRS SINCE 1952" (mirrored, seen from inside). Dozens of old clocks on the wall, all showing different times. Shallow depth of field, photorealistic, rich detail.
21
We tried the same storm prompt on Claude Opus 5.5 and Claude Fable 5.2. Every wave, raindrop and lightning bolt is pure code, no 3D models, no footage. Which one would you trust with the boat? Try both on OpenClips
1
1
5
72
Prompt 👇 Create a single self-contained HTML file with Three.js. Direct a 15-second cinematic storm at sea, one continuous camera move, built entirely in code. 0-3s: a calm sea at dusk, a lighthouse on a rocky cliff switches on its rotating beam. 3-6s: the sky darkens, heavy rain starts and the waves grow. 6-10s: a small fishing boat climbs a huge wave, spray flying off the crest. 10-12s: a lightning strike lights up the whole scene for a moment. 12-15s: the lighthouse beam finds the boat through the rain and fog and it slips behind the cliff into calmer water. Make the ocean, waves, rain, lightning, fog and light beam as realistic as you can. No external 3D models or image textures, no real brand names or logos. The animation starts automatically, lasts 15 seconds, fills the whole window at 1920x1080, and has no UI, buttons or text. Reply with the complete HTML file only.
1
28
We tried GPT-6.2 Astra and GPT-6.2 Sol on the same prompt: direct a 15-second storm at sea in pure 3D code, from a calm dusk to a lighthouse guiding a boat home ⛈️ Astra kept the boat on top of the wave. Sol built the moodier storm… and the boat went straight through the water 👀 Which one would you trust at sea? Prompt below 👇
1
6
181
Prompt 👇 Create a single self-contained HTML file with Three.js. Direct a 15-second cinematic storm at sea, one continuous camera move, built entirely in code. 0-3s: a calm sea at dusk, a lighthouse on a rocky cliff switches on its rotating beam. 3-6s: the sky darkens, heavy rain starts and the waves grow. 6-10s: a small fishing boat climbs a huge wave, spray flying off the crest. 10-12s: a lightning strike lights up the whole scene for a moment. 12-15s: the lighthouse beam finds the boat through the rain and fog and it slips behind the cliff into calmer water. Make the ocean, waves, rain, lightning, fog and light beam as realistic as you can. No external 3D models or image textures, no real brand names or logos. The animation starts automatically, lasts 15 seconds, fills the whole window at 1920x1080, and has no UI, buttons or text. Reply with the complete HTML file only.
40
Neither of them is real 👀 Two girls, two selfies, and each one swears the other is AI. Then they peel off like stickers and you see what’s underneath: one photo and one prompt. That’s it. Both made with Seedance 2.0 by ByteDance on OpenClips, from the site or straight from Claude and ChatGPT through our MCP. The full prompts are right there in the video, pause and take them. Be honest, A or B? 👇 Your first 30 credits are enough to start typing your own ⚡
1
1
9
249
We dropped four AI video models from space into a Black Friday line 🛍️ Seedance 2.5, Kling 4.0, Wan 3.0 and Gemini Omni 1.2 Flash had to fly one unbroken shot from orbit to the store doors, and one of them hit the city with a laser on the way down 👀 Which landing would you trust? Try them all on OpenClips. Prompt below 👇
1
1
3
134
Prompt 👇 One continuous 10-second shot. Photorealistic, 16:9. The camera is a drone that dives from space all the way down to a city street, one unbroken move. 0-3s: High above Earth at dawn, the curve of the planet glows and city lights shine below. The camera dives straight down through a layer of clouds. 3-6s: The camera drops over a big city at sunrise, rooftops rushing up toward it, and levels out above a wide street. 6-10s: The camera glides low along a long line of shoppers in winter coats waiting outside a department store with a huge red "BLACK FRIDAY" sign above the doors. End state: the camera hovers just above the front of the line as the glass doors slide open. Audio: rushing wind that turns into city noise, excited chatter from the line and a store bell as the doors open. Nobody speaks to the camera.
35
We tried GPT-6.2 Astra, GPT-6.2 Sol, Claude Opus 5.5 and Claude Fable 5.2 on the same prompt: direct a 15-second rocket launch, from the pad at dawn all the way to space, in pure 3D code 🚀 No video model, no edits, just code. Three of them reached Earth orbit, one is still looking for the planet 👀 Which launch would you put on the big screen? Prompt below 👇
4
9
205
Prompt 👇 Create a single self-contained HTML file with Three.js. Direct a 15-second cinematic rocket launch, one continuous camera move, built entirely in code. 0-3s: a tall multi-stage rocket stands on its launch pad at dawn, venting vapor, with the service tower beside it. 3-6s: the engines ignite, fire and huge clouds of exhaust smoke billow out across the pad. 6-10s: the rocket lifts off and climbs, the camera follows it up through a layer of clouds. 10-13s: the first stage separates and falls away while the upper stage keeps burning. 13-15s: the rocket coasts into the black of space above the curved, glowing edge of the Earth. Make the fire, smoke, clouds, atmosphere and Earth as realistic as you can. No external 3D models or image textures, no real brand names, agency names, flags or logos. The animation starts automatically, lasts 15 seconds, fills the whole window at 1920x1080, and has no UI, buttons or text. Reply with the complete HTML file only.
1
61
We tried GPT-6.2 Astra and Sol on the same prompt: build a passenger jet in 3D code, start with every part laid out on the floor, then assemble it ✈️ Sol built 151 parts and rolled it out in a full hangar. Astra kept it clean and studio-white, with the landing gear clicking in at the very last second 👀 Same prompt, zero edits. Which build wins? Prompt below 👇
1
2
91
Prompt 👇: Create a single self-contained HTML file with Three.js. Model a modern wide-body passenger airliner entirely in code, broken down into at least 60 separate components: fuselage barrel sections, cockpit nose, each windshield pane, cabin doors, wing skins, flaps, slats, ailerons, winglets, both engines split into nacelle, fan, individual fan blades, core and exhaust cone, engine pylons, horizontal stabilizers, elevators, vertical fin, rudder, tail cone, and the nose and main landing gear with every strut and wheel. The animation starts fully disassembled: every component lies neatly laid out side by side on a light grey workshop floor like an illustrated parts catalog (knolling), clearly separated, seen from above at an angle. Hold that for 2 seconds, then the parts lift and fly into place one by one in a logical build order until the finished airliner stands complete at second 10. The camera starts high above the laid-out parts and slowly moves down and around to a low three-quarter hero view of the finished plane, never passing through any object. No external 3D models or image textures, no real brand names, airline names or logos. The animation starts automatically, lasts 10 seconds, fills the whole window at 1920x1080, and has no UI, buttons or text. Make it as detailed and realistic as you can. Reply with the complete HTML file only.P
1
33
GPT-6.2 Astra vs Sol, but they’re not the ones making the video 👀 Each one wrote a prompt, Seedance 2.5 did the rest. Watch where the paper plane ends up ✈️
5
216
GPT-6.2 Astra for your hardest projects. Sol for everyday work. Both out today. Give either one a studio: connect OpenClips MCP in ChatGPT. 2 phone photos of a shoe, 66 files back. AI product videos and ads inside ChatGPT. openclips.ai/mcp
Some of you have been asking very politely. Some of you have just been asking. GPT-6.2 Astra and Sol are now available. Super efficient and built to perform where it counts—the real world. openai.com/index/introducing…
1
6
152
Smarter models writing prompts means better scenes on our side 🎬 Curious to see what GPT-6.2 Astra does with a 10-second cinematic prompt. Who’s testing it first?
Some of you have been asking very politely. Some of you have just been asking. GPT-6.2 Astra and Sol are now available. Super efficient and built to perform where it counts—the real world. openai.com/index/introducing…
2
128