Loopcut

Pricing

Bring your ChatGPT plan. Loopcut adds the rest.

Your ChatGPT Plus or Pro plan runs the agent in Blender, with no charge from us. A Loopcut account adds what a chat plan can't: asset search, images, picture to 3D, voice, sound effects and cloud render, paid from one credit balance. Every service shows its price in dollars before it runs.

The best setup is both

Neither one pays for the other.

Your ChatGPT plan: the conversation

Sign in with ChatGPT in Blender. ChatGPT Plus or Pro required; usage counts against your plan, never your credits.

Your Loopcut account: the tools

Asset search and cloud services: render, images, 3D, voice and sound. Priced in credits, shown in dollars.

No ChatGPT plan? Loopcut's own models run the agent too: Fast on every plan, the Pro model on Pro and Max.

Loopcut plans

Prices in USD. Change or cancel any time from your account.

2 months free

Free

Try Loopcut's cloud services on a real scene

$0/ month

No card needed

Up to 20 minutes of cloud GPU rendering a month, Cycles or EEVEE

On an RTX 4090. 100 credits every month ($1).

  • Works with your ChatGPT plan
  • Asset search across free libraries
  • Loopcut Fast, with a weekly allowance
  • Every feature of the agent
  • No card needed

Pro

Recommended

For artists who use cloud services every week

$20/ month

Billed monthly

Up to 7 hours of cloud GPU rendering a month, Cycles or EEVEE

On an RTX 4090. $22 of credits included every month.

Worth more than the plan costs, and good for any cloud service.

  • Everything in Free, plus:
  • The Pro model, with our Blender skills, for hard scenes and multi-step jobs
  • 7× the allowance of Free
  • Top-ups when you run low
  • Works with your ChatGPT plan

Max

For heavy render and generation months

$60/ month

Billed monthly

Up to 22 hours of cloud GPU rendering a month, Cycles or EEVEE

On an RTX 4090. $66 of credits included every month.

Worth more than the plan costs, and good for any cloud service.

  • Everything in Pro, plus:
  • 3× the allowance of Pro
  • Room for long sessions and heavy scenes
  • First to get new models

Top-ups

For a busy month on Pro or Max. Pay once; top-up credits never expire, and plan credits are spent first.

$10 of usage · 1,000 credits

With Pro or Max

$25 of usage · 2,500 credits

With Pro or Max

$50 of usage · 5,000 credits

With Pro or Max

Allowances are model time in a 5-hour session and a 7-day window. Quick Fast edits use little of it; heavy scenes, viewport captures and long Pro turns use more. The app always shows what is left.

More than a chatbot wired to Blender

Loopcut compared with Blender MCP, Claude's connector, and a chat or coding agent alone.

FeatureLoopcutMCP for Blender or Claude's Blender connectorChatGPT, Claude or Codex on their own
Runs code in your open sceneYesYesPartly. You paste scripts, or it runs Blender headless
Looks at the viewport to check its own workYesYesNo
Runs on the ChatGPT plan you already pay forYesPartly. Only through CodexYes
Every turn that changes the scene is a checkpoint. Undo all restores the scene and the conversation to before it.YesNoNo
Built-in asset search: BlenderKit, Poly Haven, Poly Pizza and your own librariesYesPartly. MCP for Blender: Poly Haven, Sketchfab and Poly PizzaNo
Picture to textured 3D modelYesPartly. MCP for Blender, with your own Tripo, Hunyuan3D, Rodin or Meshy key, or its paid planNo
Cloud remesh to clean quadsYesNoNo
Voice lines and sound effectsYesNoNo
Cloud render on an RTX 4090YesNoNo
One balance for every service, priced before it runsYesPartly. Separate keys and bills per vendorNo
Open-source agent you can read and changeYesPartly. MCP for BlenderNo

From each project's public docs, checked October 2026. MCP for Blender (formerly Blender MCP) is the open-source add-on by Siddharth Ahuja; Claude's Blender connector is the one Blender Lab built for Claude. What each Blender MCP does. They change often: if a row is out of date, tell us.

What credits buy: cloud services

One balance for everything. You see the price before every job.

The free plan's 100 credits a month buy about 20 minutes of cloud render, or 1 picture-to-3D model (or 4 drafts), or 16 images (5 in HD), or up to 20 remeshes, or 25 sound effects, or 10 minutes of voice, or 50 seconds of motion from text, or 33 seconds of motion from video. Pro's 2,200 a month buy about 7 hours of cloud render, or 27 picture-to-3D models (or 88 drafts), or 366 images (110 in HD), or up to 440 remeshes, or 550 sound effects, or 3 hours of voice, or 18 minutes of motion from text, or 12 minutes of motion from video.

ServicePriceDetails
Cloud render$0.05 per GPU-minute5 credits a minute, one-minute minimum per job. The estimate, test frame and frame checks are free. Cycles or EEVEE, on any credits.
Image$0.06 standard, $0.20 HD6 or 20 credits an image, for references and concepts. Editing an image costs the same.
Image to 3D model$0.80 a model80 credits; a quick draft is 25.
Cloud remeshFrom $0.05 a mesh5–60 credits depending on the mesh; you approve the most it can cost first. Free on your own machine.
Voice line$0.10 per minute10 credits a minute of speech, rounded up per line, 1 credit minimum.
Sound effect$0.04 a take4 credits up to 6 seconds, then 1 per extra 2 seconds.
Motion from video$0.03 per second3 credits a second of video, rounded up, 2-second minimum, up to 60 seconds a clip, captured with NVIDIA GEM-X. Putting the motion on your rig happens in Blender, free.
Motion from text$0.02 per second2 credits a second of motion, 2–10 seconds a clip, one free redo. NVIDIA Kimodo. Built with Meta Llama 3.
Asset searchFreeSearch free-licence libraries and your own.

Prices in USD; 1 credit is $0.01. Every plan includes monthly credits, and on Pro and Max top-ups add more. Work that can run on your own machine stays free.

What each model does with your data

None of them train on your requests.

ModelProviderReceivesTrains on your dataKept forProcessed in
Fast
openai/gpt-6-luna
OpenAI, through OpenRouterMessages, scene details the app chose to include and viewport images. OpenRouter receives the same content when it relays a request.NoNone at the host (OpenRouter's no-data-collection filter); OpenRouter stores noneOpenRouter in the US; OpenAI in the US
Pro
openai/gpt-6.1-sol
OpenAI, through OpenRouterMessages, scene details the app chose to include and viewport images. OpenRouter receives the same content when it relays a request.NoNone at the host (OpenRouter's no-data-collection filter); OpenRouter stores noneOpenRouter in the US; OpenAI in the US
Claude Opus 5.5
anthropic/claude-opus-5.5
Anthropic, Amazon Bedrock or Google Vertex AI, via OpenRouterMessages, scene details the app chose to include and viewport images. OpenRouter receives the same content when it relays a request.NoUp to 30 days at Anthropic; none at Bedrock or Vertex AI; OpenRouter stores noneGlobal by default; OpenRouter in the US
Backup for Fast
openai/gpt-6-luna
OpenAI, through OpenRouterMessages, scene details the app chose to include and viewport images. OpenRouter receives the same content when it relays a request.NoNone at the host (OpenRouter's no-data-collection filter); OpenRouter stores noneOpenRouter in the US; OpenAI in the US
Backup for Fast and Pro
mimo-v2.6-pro
XiaomiMessages, scene details the app chose to include and viewport images. OpenRouter receives the same content when it relays a request.NoNo fixed period statedSingapore (Xiaomi processes API requests by the account holder's region; Loopcut's account is outside the EEA)
Backup for Fast and Pro
xiaomi/mimo-v2.6-pro
A MiMo-V2.6-Pro host chosen by OpenRouterMessages, scene details the app chose to include and viewport images. OpenRouter receives the same content when it relays a request.NoNone at the host (OpenRouter's no-data-collection filter); OpenRouter stores noneOpenRouter in the US; the host's region varies

From each provider’s published terms, last checked 9 October 2026; the answer in each row links to them. “Not stated” means the terms give no figure. A backup only answers when the model it backs up is busy, and you are charged at the backup’s price.

With your ChatGPT plan, own API key or a local model, Loopcut’s gateway is not involved in model requests. OpenAI’s or your configured provider’s terms apply to requests sent to them.

Your ChatGPT plan, your own key, or a local model

Free, forever. No Loopcut plan needed.

Sign in with ChatGPT and the agent runs on your Plus or Pro plan. Or paste a key from your provider, or the address of a model on your own machine. Requests go straight from Blender to the provider you chose, never through our server. The app still fetches our service list, sends Coming soon clicks and, from version 0.2.6 unless you turn sharing off, a log of each conversation, as described on /privacy. Set it up in Preferences › Add-ons › Loopcut.

  • Your ChatGPT Plus or Pro plan, signed in from Blender
  • Any OpenAI-compatible API, including local servers
  • Every feature of the agent, no limits from us
  • Download Loopcut

Questions

Straight answers.

Anything else? Open an issue and we will answer in public.

What is an allowance?
Each plan gives you an amount of model time in two windows: a 5-hour session and a 7-day week. Model time is the work the model does for you, so a long Pro turn on a heavy scene uses more than a quick Fast rename. The app shows how much of each window is left.
What counts as a turn?
One message to Loopcut with a few tool steps: reading the scene, writing and running the code, checking the viewport. Heavy scenes and viewport captures send more context, so they cost more.
What happens when I run out?
Requests pause until the window resets, and the app tells you when that is. Nothing is deleted and you are never charged extra. If you hit the limit often, the next plan up is the fix.
How is Loopcut different from Blender MCP or Claude's Blender connector?
Those connect a chat app to Blender, so the chatbot can run code in your scene. Loopcut is the agent itself, built into Blender: it checkpoints every step, searches asset libraries, and runs cloud services (3D models, remesh, voice, sound and render) from one credit balance, with the price shown before each job. The conversation can run on the ChatGPT plan you already pay for.
Do I need both a ChatGPT plan and a Loopcut account?
No. Either works alone. With only your ChatGPT plan, the agent runs on it and every free tool works. With only a Loopcut account, the agent runs on Loopcut Fast. Together you get the model you already pay for plus the cloud services a chat plan can't run.
Do credits pay for my ChatGPT usage?
No. On your ChatGPT plan the conversation goes from Blender to OpenAI and counts against your plan's own limits. Credits pay for GPUs and services only: render, images, 3D, voice and sound.
Do I need a paid plan at all?
No. Sign in with ChatGPT and the agent runs on your Plus or Pro plan; or paste a key from any OpenAI-compatible provider, or point Loopcut at a local model. All of these are free with no Loopcut plan. Paid plans add more monthly credits, top-ups and the Pro model, and they fund the project.
Can I use my own vendor keys instead of credits?
Yes. Add an OpenAI, fal, ElevenLabs or Gemini key in Preferences › Add-ons › Loopcut, and the tools that use it run on your key with no charge from us.
What do you keep from my sessions?
With your ChatGPT plan, own key or a local model, model requests go straight from Blender to the provider, never through our server. From version 0.2.6 the app sends us a log of each conversation (prompts, replies, tool calls and errors, never images or keys) unless you turn sharing off under Preferences › Add-ons › Loopcut; we keep it for 90 days to help you and fix problems, and never use it to train AI models. The app also fetches our service list and sends Coming soon clicks, as described on /privacy. On hosted plans the gateway relays your request to the model provider and records token counts, cost and timing for metering. The gateway itself does not store your prompts or scene data. The provider's own terms cover what they receive; the table above lists each provider and whether it trains on your data.
Can I cancel or change plans?
Any time, from the account page. Downgrades and cancellations take effect at the end of the billing period; upgrades apply straight away.
How does yearly billing work?
Pro and Max can be paid yearly: 12 months for the price of 10. The plan's credits still arrive every month and last until the next ones arrive, as on monthly billing. Monthly stays the default, and the same 14-day refund rule applies.
How do cloud services work?
Cloud render and the other cloud services run on our machines and are paid from one balance of credits. A credit is one cent: Loopcut shows the price in dollars before every job, and a job never costs more than the price you approved. Failed jobs are refunded automatically.
Do credits expire?
Monthly credits last until the next ones arrive; unused monthly credits don't carry over. A paid plan's arrive at the start of each billing month. The free plan's 100 arrive at the start of each calendar month (UTC); the 100 you get when you sign up are your first month's, and they stay until you use them. Top-up credits stay until you use them too, and we spend monthly credits first. There is no overage and no automatic top-up: when the balance runs out, cloud services pause and the agent keeps working.
Can I buy more credits?
On Pro or Max, yes: top up with $10, $25 or $50 of credits from this page or your account page, as often as you need. Top-ups are for paid plans; on the free plan, the next step is Pro.
Can I render with Cycles on the free plan?
Yes. Cycles and EEVEE cloud renders run on any credits, including the free plan's 100 a month: up to 20 GPU-minutes on an RTX 4090. They also cover every other cloud service.
Is Loopcut the same as Blender's Loop Cut tool?
No. Loop Cut and Slide (Ctrl+R) is Blender's own tool for adding edge loops, and it works as before. Loopcut is a separate, open-source AI agent that ships inside its own Blender build.

Not sure yet? Start free.

A free account gives you asset search, 100 credits every month for cloud services, and Loopcut Fast. No card needed.