
1 / 3
【Overview】 Clusy is an AI notebook IDE for Jupyter built around an autonomous agent. Instead of writing every cell yourself, you describe the goal in plain language and the agent plans the work, writes and runs notebook cells on managed cloud CPUs or GPUs, and fixes what breaks along the way. The premise is that a one-sentence request can become experiments, models and answers without you babysitting the kernel. The product targets the messy middle of computational work: not just writing code, but running several ideas, comparing them, and scaling up when something works. Branching lets three approaches run at once, each on its own branch and kernel, so you can compare results side by side rather than one after another. When a run looks promising, you can snapshot the whole workspace, move it to a bigger GPU and resume from the checkpoint. Clusy is aimed at data teams, researchers, engineers and data scientists who already work in notebooks and want the agentic layer plus managed compute without assembling the infrastructure themselves. Existing .ipynb files can be imported, and finished notebooks can be exported back or published to Hugging Face, Kaggle or GitHub. 【Key features】 • Autonomous agent: Plans the work, writes and runs notebook cells from a plain-language request, and repairs failures. • Parallel experiments with branching: Runs three approaches at once, each on its own branch and kernel, for side-by-side comparison. • Snapshot & upscale: Moves the entire workspace to a larger GPU and resumes from the checkpoint. • Managed sandboxes: Runs notebooks on cloud sandboxes from CPU-only up to T4, L4, A10, L40S, A100, H100 and H200 GPUs. • Data source connections: Connects uploaded files, public Hugging Face and Kaggle datasets, and warehouses such as Snowflake and Databricks. • Model choice: Selects from Auto, GPT-6 Luna, Haiku, DeepSeek, open models (Kimi, Qwen, GLM), Claude and GPT depending on plan. • Notebook import/export: Brings existing .ipynb files into a Clusy notebook and exports them back to .ipynb. • Publish and share: Publishes to Hugging Face, Kaggle or GitHub in one click, or turns any run into a stable shareable link others can open and fork. 【Use cases】 1. Train a chargeback model on company data and choose a threshold within budget. 2. Fine-tune LLMs end-to-end, including dataset discovery, preprocessing and sandboxed training. 3. Run parallel modeling experiments and compare results across branches. 4. Reproduce a model from a paper and run experiments. 5. Idea validation and market research. 【Pricing】 Free: $0/month — Auto model for everyday agentic work, 5 GB of checkpoint storage, CPU sandbox with 8 vCPU and 8 GB RAM; checkpoints released after 3 days. Plus: $12/month — GPT-6 Luna, Haiku and DeepSeek, 30 GB of checkpoint storage, T4 GPU (16 GB VRAM), up to 16 GB RAM, indefinite retention. Pro: $30/month — 3x Plus usage, adds all open models (Kimi, Qwen and GLM), 100 GB of checkpoint storage, all entry GPUs (T4, L4, A10) and 32 GB RAM. Max 10x: $90/month — 10x Plus usage, every model including Claude and GPT, 512 GB of checkpoint storage, GPUs up to the A100 and 64 GB RAM. Max 30x: $200/month — 30x Plus usage, every model including Claude and GPT, 2 TB of checkpoint storage, GPUs up to the H200 and 128 GB RAM. 【Pros and cons】 Pros: - The free tier is genuinely usable for agentic notebook work on CPU, so you can judge the agent's planning quality before paying. - Branching plus snapshot-and-upscale suits exploratory work: try several cheap approaches, then commit compute only to the one that looks best. - The price ladder scales on usage and hardware rather than seats, which fits individual researchers and small data teams better than per-user licensing. Cons: - GPU access is gated by plan: the Free tier is CPU only, and the H200 with up to 141 GB VRAM is available only on Max 30x. - Claude and GPT models are limited to the Max plans. - On the Free plan, checkpoints are released after 3 days, which makes it unsuitable for long-lived work. 【FAQ】 Q: Can I try Clusy without paying? A: Yes. The Free plan is $0/month and includes the Auto model, a CPU sandbox with 8 vCPU and 8 GB RAM, and 5 GB of checkpoint storage, with checkpoints released after 3 days. Q: Which plan do I need for Claude or GPT models? A: Claude and GPT are available only on the Max plans, starting with Max 10x at $90/month. Q: Who is Clusy designed for? A: Data teams, researchers, engineers and data scientists who work in Jupyter notebooks and want an agent to plan, run and branch their experiments on managed cloud compute.
Free · $0/month (Auto model for everyday agentic work, 5 GB of checkpoint storage, CPU sandbox with 8 vCPU and 8 GB RAM) Plus · $12/month (GPT-6 Luna, Haiku and DeepSeek, 30 GB of checkpoint storage, T4 GPU (16 GB VRAM), up to 16 GB RAM) Pro · $30/month (3x Plus usage, Adds all open models: Kimi, Qwen and GLM, 100 GB of checkpoint storage) Max 10x · $90/month (10x Plus usage, Every model, including Claude and GPT, 512 GB of checkpoint storage) Max 30x · $200/month (30x Plus usage, Every model, i
WPS AI is an AI assistant built into WPS Office that helps users write documents, generate presentation outlines and…
Baidu Comate is an AI coding assistant built by Baidu on its ERNIE model. It goes beyond simple autocomplete: it…
UniConverter is an AI media conversion and enhancement tool for video, audio, and images. It combines everyday…
6 similar tools to consider instead of Clusy