Content Studio, with Runway built in
Your growth team has its own film studio.
Content Studio turns one brief into Runway video clips, pictures, voiceovers, music and sound effects. Your @runners pick the right AI model for every piece, check it against your brand, and bring it to you to approve.
Every format your channels ask for
Write the brief once: the objective, the audience, the offer and the guardrails. Every take starts from it, in whatever shape the channel wants.
Video clips
Runway turns a sentence or a single still into a clip of up to 15 seconds: vertical 9:16, the shape Reels, TikTok and Shorts use, or square, portrait and widescreen. Seedance 2.5 on Higgsfield runs one shot up to 30.
Pictures
Post pictures and ad creatives from GPT Image or Runway's Gen-4 Image, which works from up to three reference pictures of the real thing.
Voiceovers
A multilingual ElevenLabs voice reads your script, with captions timed to match.
Music and sound effects
Instrumental music beds from three seconds to a minute, and sound effects from a whoosh to a downpour, looped when you need it.
Presenter videos
A presenter speaks your script to camera, marked as AI, framed vertical, square, portrait or widescreen.
The final cut
Join up to 24 shots into one film with an end card, then mix a music score and sound effects underneath.
Every clip, voiceover, music bed and sound effect you make lands in the Library, searchable by name, words or tag. Reuse a music bed or a voiceover under another clip at no charge.
Runway, built in
Runway's video models, inside your studio
Connect Runway (RunwayML) and your team makes text-to-video and image-to-video clips on the models below, without leaving Runner.
- Gen-4 Turbo
- The fast standard. Brings a still to life in up to 10 seconds: vertical, square, portrait or widescreen.
- Gen-4.5
- Text to video or image to video, from one sentence or one still.
- Veo 3.1
- Google's model, for a shot that starts and ends on the frames you choose.
- Seedance 2
- Up to 15 seconds, guided by as many as nine reference pictures and a reference clip.
- Gen-4 Image
- Runway stills from a prompt and up to three reference pictures.
- The model fits the shot: words alone go to Gen-4.5, a first and last frame to Veo 3.1, reference pictures to Seedance 2.
- Best of two: make the same clip on a second Runway model and keep the one you'd post.
- A clip made from words or frames that Runway's moderation turns down is retried once on the other model family, on its own.
- Use your own Runway key, or make clips on AI credits: a 10-second clip on the standard model is 100 credits, and you see the price before it's made.
Higgsfield, for the longer take
Connect your Higgsfield key and your team makes clips on two more models, without leaving Runner. Higgsfield bills your account at its own prices.
- Seedance 2.5
- One shot of up to 30 seconds, from words or a still. In your picker it's Long take.
- MiniMax H3
- Up to 15 seconds in 2K, from words or a still.
Claude writes it. Runway shoots it. ElevenLabs voices it.
Or any mix you like. Runner connects to the AI models below for writing, pictures, video and sound, and sends each job to the one it suits. Bring your own keys, or run on AI included and skip the keys altogether.
Runway
Video and stills
- Gen-4 Turbo
- Gen-4.5
- Veo 3.1
- Seedance 2
- Gen-4 Image
Higgsfield
Longer video, on your own key
- Seedance 2.5
- MiniMax H3
OpenAI
Pictures and writing
- GPT Image 2
- GPT-6.1 Sol
- GPT-6.1 Terra
ElevenLabs
Voice, music and sound
- Multilingual voices
- Music beds
- Sound effects
Anthropic
Writing and planning
- Claude Fable 5.1
- Claude Opus 5.5
- Claude Sonnet 5.5
- Claude Haiku 5.5
Google
Writing
- Gemini 3.1 Pro
xAI
Writing
- Grok 4.7
Ollama
Writing, on your own key
- DeepSeek V4.1 Flash
- GLM-5.3 Flash
- Kimi K3
- Gemma 4
Runway, Gen-4, Veo, Seedance, Higgsfield, MiniMax, OpenAI, GPT, ElevenLabs, Anthropic, Claude, Google, Gemini, Gemma, xAI, Grok, Ollama, DeepSeek, GLM, Moonshot AI and Kimi are trademarks of their owners. Runner is not affiliated with any of them.
Send every job to the right model
Runner's model router decides which AI model makes each piece: per shot, per channel or per product. Start from one of these setups, mix them, or let the team work out its own.
Tryouts
DefaultLet the team decide
The team tries your stills and clips on each model the product can use, learns what your audience rewards on each channel, and backs the winner. One piece in five goes to the runner-up, so a better model still gets its shot.
Set it in Settings › Pictures and video › Let the team decide
Channel Captain
A model for every channel
Switch off Let the team decide and give Instagram, your Facebook Page, LinkedIn, Threads and your Reddit, Meta, LinkedIn and Google ads each their own picture and video model. Once a week the team proposes a switch when the results favour one.
Set it in Settings › Pictures and video › Per channel
House Style
One look across the product
Switch off Let the team decide and set one picture model and one video model for the product, so every brief looks like it came from the same studio. A channel route, or a pick for a single piece, still wins.
Set it in Settings › Pictures and video › Images and Video
Director's Cut
The right Runway model for each shot
Gen-4.5 from words, Gen-4 Turbo from a still, Veo 3.1 when you set the first and last frame, Seedance 2 when you hand it reference pictures.
Set it in Settings › Pictures and video › Video › Runway: Auto
Speed Run
Fast and light
Haiku 5.5 on the channel leads' drafts and Gen-4 Turbo on every clip made from a still: the quickest turnaround, at the lowest clip price.
Set it in Growth team › Tune a channel lead › Faster
Brain Trust
Heavy thinking where it counts
Fable 5.1 or Opus 5.5 at Deepest thinking for the chief executive who writes the team's plan, lighter models on the channel leads' everyday drafts. Bigger models cost more per draft, so spend them where the thinking pays.
Set it in Growth team › Tune the chief executive › More capable
When setups overlap, the closest one wins: the piece itself, then its channel, then the product, then a creative skill, then Auto. With Let the team decide on, your channel and product picks are where the team starts. Auto names the model it will use and where that choice came from.
Your rules go into every take, and you sign off each one
1. Your rules, built in
Your imagery rules go into every take, and a take that cites a claim you've ruled out is flagged before you approve it.
2. Never means never
Tell it what never to show. A prompt that asks for it is refused before any model is paid, and a vision model checks every picture against the list before it's kept.
3. Six checks, one signature
Technical, brand, claims, rights, policy and client. Approving a take signs all six under your name, and you choose where it may run.
4. A record of every shot
The debrief puts every shot, prompt, model and credit on one page you can print or save as a PDF.
Questions about Content Studio
Does Runner work with Runway (RunwayML)?
Yes. Runway is built into Content Studio. Your team makes clips on Gen-4 Turbo, Gen-4.5, Veo 3.1 and Seedance 2, and makes stills with Gen-4 Image. Connect your own Runway API key, or make clips on AI credits.
Does Runner work with Higgsfield?
Yes. Connect your Higgsfield key in Content Studio and your team makes clips from a sentence or a still on Seedance 2.5, up to 30 seconds, and MiniMax H3, up to 15 seconds in 2K. Higgsfield bills your account at its own prices.
Which AI video generators can I use?
Runway's Gen-4 Turbo and Gen-4.5, Google's Veo 3.1 and ByteDance's Seedance 2, all through Runway, and ByteDance's Seedance 2.5 and MiniMax H3 through Higgsfield on your own key, for text to video and image to video. Clips run up to 15 seconds, or up to 30 on Seedance 2.5, vertical 9:16, square, portrait or widescreen. Pick one for a clip, set one for the product, or let the team choose.
Which AI image generators does it use?
OpenAI's GPT Image and Runway's Gen-4 Image, which works from up to three reference pictures. Drawing an Instagram post picture, you can pick GPT Image 2, 1.5, 1 or 1 mini. Every picture is checked against your list of things never to show before it's kept.
Can it make voiceovers, music and sound effects?
Yes, with ElevenLabs. A multilingual voice reads your script with captions to match, labelled AI voice. Music beds are instrumental, three seconds to a minute, and carry Content Credentials. Sound effects run from half a second to thirty and can loop. Ask for the same voiceover or music again and you can reuse it at no charge.
What is an AI model router?
A model router sends each job to the AI model best suited to it, rather than running everything through one. In Runner it works per piece, per channel and per product, and Auto always shows which model it picked and why. With Let the team decide on, the team learns which model your audience responds to on each channel.
Can I use a different AI model for each channel?
Yes. Give Instagram, your Facebook Page, LinkedIn, Threads and the Reddit, Meta, LinkedIn and Google ad networks each their own picture and video model, or leave a channel on the product default. With Let the team decide on, those picks are where the team starts. Switch it off and your routes stand, and the team proposes a switch once a week when the results favour one.
Which AI models write the posts and scripts?
Your growth team writes with Claude (Fable 5.1, Opus 5.5, Sonnet 5.5 or Haiku 5.5), OpenAI's GPT-6.1 Sol and Terra, Google's Gemini 3.1 Pro or xAI's Grok 4.7. You choose the model for the chief executive and each channel lead, and how hard it thinks.
How much does a Runway clip cost?
On your own Runway key, Runway bills your account at its own prices. On AI included, a 10-second clip is 100 credits on the standard model and 240 to 720 on a premium one, and each clip shows its credits before it's made.
Is AI-made content labelled?
Yes. You see which model made each picture or clip, and every picture a model draws carries a machine-readable mark naming that model. Voiceovers are labelled AI voice, presenters are marked as AI, and music carries Content Credentials.
Does anything go out without me?
Not unless you say so. Every take in Content Studio waits for a person to approve it and choose where it may run, and approving signs off its six checks under your name. Your growth team posts without asking only where you've saved a rule that lets it, inside the limits you set.
Put your team behind the camera.
Write one brief and your @runners take it from script to screen: Runway clips, pictures, voice and music, ready for you to approve.