Media & research keys

Some LoopCodeLab features call outside providers, so they need a key that belongs to you. This page covers the keys for generating video, images, and audio, for grounded research and real datasets, for stock imagery, and for the live-edit sandbox. Each one is optional, and you only connect the ones you want to use.

You add every key the same way. In LoopCodeLab, open Settings, find the provider’s card, paste the key, and save. Each card has a Test button that runs a quick sign-in check without spending anything, so you can confirm a key works before you rely on it in a build.

Prices below are current as of July 2026 and change often, so always confirm the number on the provider’s own pricing page before you commit.

Video, image and audio generation

Media generation is opt-in for every build, and you pick which kinds you want (image, video, audio) when you set the build up. Each kind has a hard cap, and the planned amount is shown before you start, so there are no surprise runs.

A media build will not start unless the matching provider key is connected:

• Music needs a Suno key.
• Voiceover needs an ElevenLabs key or the Grok sign-in.
• Image or video needs an image or video provider (any of the ones below).

BytePlus ModelArk / Seedance (video)

What it unlocks: AI video generation for video builds and video scenes in social builds.

How to get the key: BytePlus uses one ARK key for your whole account, and it is the same key the GLM coding agent uses, so if you have already connected GLM you are done. To get a new one, sign in at byteplus.com and create an ARK API key in the ModelArk console.

Billing note: the /api/v3 video endpoint bills pay-as-you-go and is billed separately from any coding plan, so video spend does not draw down GLM coding credits.

Current price: Seedance video is billed by tokens, which depend on duration, resolution, and frame rate. BytePlus lists Seedance 2.0 at roughly $4.3 per 1M tokens with a video input and $7 per 1M tokens without one at 720p, with 1080p at $4.7 / $7.7 and a faster tier at $3.3 / $5.6 (BytePlus ModelArk pricing, as of July 2026). Converted to a per-second figure, third-party estimates put the lowest tier near $0.045 per second of video (EvoLink Seedance 2.0, as of July 2026).

Base URL: https://ark.ap-southeast.bytepluses.com/api/v3

Adding it: paste the ARK key into the BytePlus card in Settings and press Test to confirm it authenticates.

Grok Imagine (image and video)

What it unlocks: image and video generation, and it also counts as a voiceover provider.

How to get the key: Grok Imagine needs no separate key. It runs on the same Grok (xAI) subscription sign-in you use for the Grok coding agent, using that plan’s Imagine credits. If you have signed in to Grok in Settings, Imagine is already available.

Current price: there is no extra media bill beyond your Grok (SuperGrok) subscription. For the plan cost, see the provider’s pricing page at x.ai.

Base URL: xAI’s Imagine models on api.x.ai (handled for you through the Grok sign-in).

Adding it: use the Grok card’s browser sign-in in Settings. Once you are signed in, Grok is available as an image, video, and voiceover provider with no key to paste.

Token-plan image (Qwen MaaS)

What it unlocks: image generation through Alibaba’s MaaS image models.

How to get the key: there is no separate key. It reuses your Qwen Alibaba token-plan key, so if Qwen is connected for coding, image generation over this path works too.

Current price: covered by your Qwen token plan. See the provider’s pricing page for the plan you hold.

Adding it: nothing extra to add. Connect the Qwen token-plan key in Settings and this image path comes with it.

Suno (music)

What it unlocks: music generation for audio builds (songs, with lyrics you provide or lyrics the agent writes when needed).

How to get the key: create an account and an API key at sunoapi.org.

Current price: sunoapi.org sells credits, at about $5 for 1,000 credits (roughly $0.005 per credit) on the base tier, falling to about $0.00455 per credit at higher volume (Suno API pricing 2026, Sunor, as of July 2026). Confirm the current tiers on the sunoapi.org pricing page, since third-party Suno pricing moves around.

Base URL: https://api.sunoapi.org

Adding it: paste the Suno key into the Suno card in Settings and press Test.

ElevenLabs (voiceover)

What it unlocks: voiceover generation for audio builds and narrated video.

How to get the key: sign in at elevenlabs.io, open your profile, and copy your API key from the API key section.

Current price: ElevenLabs has a free plan with about 10,000 credits a month, then Starter at $5 a month for 30,000 credits, Creator at $22 a month for 121,000 credits, and Pro at $99 a month for 500,000 credits. On the standard Multilingual v2 model one credit is one character, and the Flash and Turbo models use half a credit per character (ElevenLabs pricing 2026, Smallest.ai, as of July 2026). Confirm on elevenlabs.io/pricing.

Base URL: https://api.elevenlabs.io/v1

Adding it: paste the ElevenLabs key into its card in Settings and press Test. If you would rather not connect it, the Grok sign-in also covers voiceover.

Research and data

Perplexity and Apify are optional. They only improve planning and data gathering, and a build runs fine without them. Connect them when you want the extra grounding.

Perplexity (grounded planning research)

What it unlocks: live-web grounding during planning, so your build’s plan can reference the current world and cite sources. Workers can also pull cited answers during the build.

How to get the key: create an API key in your account settings at perplexity.ai.

Current price: the Sonar model is $1 per 1M input tokens and $1 per 1M output tokens, plus a per-request search fee of $5, $8, or $12 per 1,000 requests for low, medium, or high search context (Perplexity pricing, as of July 2026). LoopCodeLab uses the low context setting, so grounding lands near the $5 per 1,000 requests figure.

Base URL: https://api.perplexity.ai

Adding it: paste the Perplexity key into its card in Settings and press Test. The test is a small ping, so it costs almost nothing.

Apify (real datasets)

What it unlocks: real datasets pulled into a build, so workers can scrape or fetch live data instead of making it up.

How to get the key: sign in at apify.com, then open Settings and the Integrations section to copy your API token.

Current price: the free tier gives $5 in credits each month with no card required, and paid plans start at Starter for about $29 a month (Apify pricing 2026, as of July 2026). Your real cost depends on compute units, proxies, and storage, so scraping actors spend from your credit budget. Confirm on apify.com/pricing.

Base URL: https://api.apify.com/v2

Adding it: paste the Apify token into its card in Settings and press Test.

Stock images

Stock keys are optional. Without them, LoopCodeLab falls back to keyless placeholder imagery, so builds still work. Add a key when you want real stock photos instead. Both providers below are free.

Unsplash (free stock images)

What it unlocks: free stock photography for build imagery.

How to get the key: register an application at unsplash.com/developers and copy the Access Key.

Current price: free.

Base URL: api.unsplash.com

Adding it: paste the Unsplash access key into its card in Settings and press Test.

Pexels (free stock images and video)

What it unlocks: free stock images and video clips for build imagery.

How to get the key: request a free API key at pexels.com/api.

Current price: free.

Base URL: api.pexels.com

Adding it: paste the Pexels key into its card in Settings and press Test.

Live-edit sandbox (Daytona)

Daytona

What it unlocks: the Visual Edit “Go live” feature, which runs a finished web app in a short-lived cloud sandbox so you can see edits render before you keep them. Without a Daytona key, edits still work using the on-box static preview, so this key only adds the live renderer.

How to get the key: sign up at daytona.io and create an API key in the dashboard.

Current price: Daytona includes $200 in free compute and 5 GB of free storage with no card for the trial. After that it bills per second, starting around $0.0504 per vCPU hour, $0.0162 per GiB of memory per hour, and $0.000108 per GiB of storage per hour, which works out to roughly $0.067 an hour for a small 1 vCPU, 1 GiB sandbox while it runs. When a sandbox is stopped you pay only for storage (Daytona pricing, as of July 2026).

Base URL: https://app.daytona.io/api

Adding it: paste the Daytona key into its card in Settings and press Test. The sandbox is ephemeral and idles out on its own, so it only runs while you are actively editing.