Flowith is an AI workspace where, instead of stacking messages in a chat box, you work by connecting nodes on an infinitely expanding canvas. Because you can branch from a single question in several directions and dig into only the ones that matter, it fits work that naturally forks, such as research and ideation. It goes beyond text generation: you can switch between more than 40 models on the canvas — including OpenAI’s GPT Image 2, Google Gemini, and the video model Kling — to produce images, video, slides, and web pages in the same workspace. It also has agent features that take a single-sentence goal, break it into tasks, and run them, which places it apart from a plain chat tool. Flowith launched in 2024 and is available on the web as well as through an iOS app.
Key Features
- Visualize your thinking on an infinite canvas: Responses are placed as nodes and connected with lines. Unlike a chat log that scrolls by in chronological order, you can see at a glance which conclusion came from which premise. Dead-end branches can stay where they are while you extend a different one
- Switch among 40+ AI models: Choose a model per node depending on the job. Text with one model, images with another — all in a single screen, so you don’t have to move between several services
- Automatic task breakdown by agent: Hand it only a goal, such as “research this and turn it into a report,” and it assembles the necessary steps and runs them in order. Intermediate results remain on the canvas, so you can check afterwards where things went off course
- Image, video, slide, and web page generation: Non-text deliverables are produced inside the same workspace. A batch mode generates multiple image and video variations at once, and higher plans allow more concurrent generations
- A single credit pool for billing: Text, images, and video all draw from the same monthly credit allocation. Credits are consumed per processing step, so heavier work burns through them faster
- iOS app support: Drop an idea onto the canvas while you are out, then organize it later on the desktop
Pricing
| Plan | Monthly | Annual | What’s included |
|---|---|---|---|
| Starter | Free | ─ | 300 credits (one-time grant), limited model access |
| Pro | $19.90 | $214.92 (about $17.91/month) | 20,000 credits, 40+ models, image and video batch mode (2x), limited concurrent tasks |
| Ultimate | $49.90 | $538.92 (about $44.91/month) | 50,000 credits, unlimited pack, batch mode (4x), 50 concurrent tasks |
| Infinite | $499.90 | $4,799.04 (about $399.92/month) | 500,000 credits, unlimited pack with all models, unlimited concurrent tasks |
Annual billing gives roughly a 10% discount on every tier, and all paid tiers include a commercial-use license. Note that the official site sometimes displays limited-time promotional prices, so check the actual amount on the checkout screen.
Pricing is current as of August 2026. Check the official site for the latest rates.
Pros & Cons
✅ Pros
- Branches of thought stay visible on screen, which suits research and planning where the conclusion is never a straight line
- Text, images, video, and slides are handled in one environment, cutting down on copy-and-paste across services
- Because models are chosen per node, it is easy to use a lighter model for drafts and a stronger one for the final pass
- You can move between handing a goal to the agent and building nodes by hand on the same canvas
- The free plan lets you try the canvas before deciding on a paid tier
⚠️ Cons
- With credit-based billing, generating many images or videos can exhaust your allocation sooner than expected
- Free-plan credits are a one-time grant, so it is not suited to ongoing free use
- A canvas-style UI can be disorienting for anyone used to chat interfaces, and takes time to get comfortable with
- The more nodes you add, the messier the canvas becomes, which adds cleanup work
- The top-tier Infinite plan is expensive for personal use and needs a careful cost-benefit check
Comparison with Similar Services
| Criteria | Flowith | ChatGPT | Manus | Miro |
|---|---|---|---|---|
| Main interface | Infinite canvas (connected nodes) | Chat + Canvas | Chat + task execution view | Online whiteboard |
| Main use | Research, production, agent execution | General-purpose chat and writing | Autonomous agent task delegation | Team diagramming and co-editing |
| Model selection | 40+ models, chosen per node | Choose among the provider’s models | Selected automatically internally | AI features are supplementary |
| Image/video generation | Supported (with batch mode) | Images supported | Limited as deliverables | Limited |
| Free plan | Yes (one-time credits) | Yes | Yes (credit-based) | Yes |
Who Is It For
- People whose research or planning branches easily, to the point that a chat log no longer captures how they got there
- People who want to produce text, images, and video together without subscribing to several separate services
- People who want to compare multiple AI models and use each one where it fits best
- People who want to try agent-style work — handing over a goal — while still watching the process
- People who already organize their thinking with whiteboards or mind maps
Summary
Flowith treats an exchange with AI as a structure on a canvas rather than a conversation history, which lets the branches of your thinking stay part of the work itself. Using 40+ models from a single screen, and combining agent features with manual node building, is what sets it apart from chat-style tools. That said, credit consumption varies widely with how you use it, so the practical approach is to start with the free Starter plan, get a feel for the canvas and your own workload, and then consider Pro or above.