diff --git a/website/src/assets/jan-vision.png b/website/src/assets/jan-vision.png new file mode 100644 index 000000000..5ba34cb13 Binary files /dev/null and b/website/src/assets/jan-vision.png differ diff --git a/website/src/content/products/index.mdx b/website/src/content/products/index.mdx index b097b2755..5c34c1165 100644 --- a/website/src/content/products/index.mdx +++ b/website/src/content/products/index.mdx @@ -13,112 +13,81 @@ import { Aside, Card, CardGrid } from '@astrojs/starlight/components'; We know it's hard but we believe this will be possible in the next decade through a combination of models, applications and tools. For this we are... -> **building a solution that ties all of these seamlessly so that users, regardless of their technical -background, are able to use Jan in the same way they use other applications but while owning them.** +> **building Jan as the ecosystem that ties all of these seamlessly so that users, regardless of their technical +background, add intelligence to their day-to-day lives like they would, but better, with similar tools.** -## What We're Building +![Jan Vision](../../assets/jan-vision.png) -**Jan Ecosystem** = Jan Models + Jan Application + Jan Tools -Unlike other AI assistants that do specific tasks with one model or have many models with a myriad of solutions, Jan provides: -- Its own specialised models that are optimised at specific tasks like web-search, creative writing, and translation -- Applications that work across all of your devices in an integrated way -- Tools to help you get things done +## Core Principles -## Two Modes +### 1) Build the Full Stack -### Local Mode -Run AI models entirely on your device, giving you complete privacy with no internet required. +Models alone aren't enough. Neither are applications. Superintelligence requires models that +understand your needs, tools that extend capabilities, and applications that tie it all +together. We're building all three, openly. -![Jan Desktop](../../assets/jan_desktop.png) +### 2) You Choose Who Runs It +Run Jan on your laptop. Self-host it on your servers. Use our cloud. The same superintelligence +works everywhere. Your data, your compute, your choice. -### Cloud Mode -Connect to more powerful models when needed - either self-hosted or via jan.ai. +### 3) Start Simple, Scale Infinitely +Open Jan and start chatting. No setup required. When you need more - better models, advanced tools, +team deployment - everything's there. The complexity scales with your ambition, not our architecture. -![Jan Everywhere](../../assets/jan_everywhere.png) +## The Path to Superintelligence - +### Today we have the Foundation -## Our Product Principles +- **A desktop app** that works both with local and cloud-based models +- **Jan models** small enough to run on any laptop and powerful enough to scale on any server +- **Basic tools** enabled through MCP Search, file parsing, simple workflows -### 1) It Just Works -1. Open Jan, start chatting -2. Onboarding is fully available but optional -3. Setting up an API key is optional -4. Selecting a local model is optional -5. Become a power user at your own pace, if you want to +### Next 12 Months: Ecosystem +- **The Jan v1 models** are a specialized series of models with general capabilities but optimsed +for specific tasks like search, analysis, creative writing and more +- **The Jan server** works as a self-hosted AI infrastructure for teams +- **Advanced tools** like browser use, deep research, and long-term memory works across devices, excels +across different day-to-day use cases, and scales with the needs of large teams +- **Cross-device sync** allows you to take your AI everywhere -We handle the complexity. +### End State: Open Superintelligence +Not one massive model, but an ecosystem of specialized models, tools, and applications working +together. Built in public. Owned by whoever runs it. -### 2) Cloud When Needed -Start completely locally and own your AI models. Add cloud capabilities only when you choose to. +## Why This Matters: The Status Quo -### 3) Solve Problems, Not Settings -We help users get to answers quickly and leave the configurations as optional. Power users can dig deeper, but it's never required. +Every other AI company wants to be your AI provider. We want you to own your AI. -## Available on Every Device +- **OpenAI/Anthropic**: Their models, their servers, their rules +- **Open Source Models**: Powerful but fragmented - no cohesive experience +- **Jan**: Complete ecosystem you can own, modify, and deploy however you want - - - **Available Now** +## Watch Us Build - Your personal AI workstation that helps with your use cases and powers other devices. Run models locally right away or bring an API key to connect to your favorite cloud-based models. +### Live Model Training +We train our models in public. Check the [models page](./models/jan-v1) to see: +- Real-time training progress +- Failed runs and what went wrong +- Models in testing before release - **[Learn more →](./platforms/desktop)** - - - **Beta Launch Soon** +No black box. No "trust us, it's good." Watch the entire process from dataset to deployment. - Web-based version of Jan with no setup required. The same default cloud mode for mobile and desktop users. +### Help Evaluate Our Models +Every model needs real-world testing. Join our open evaluation platform where you can: +- Compare model outputs side-by-side +- Test specific capabilities you care about +- Vote on which responses actually help +- Suggest improvements based on your use cases - **[Learn more →](./platforms/jan-ai)** - - - **Coming Q4 2025** +Think LMArena, but you can see all the data, run your own evals, and directly influence what we train next. - Connect to your Desktop/Server, or run local models like Jan Nano. The same experience, everywhere. +[Test/evaluate our models here](link) - **[Learn more →](./platforms/mobile)** - - - **Coming Q3 2025** +## Get Involved - Self-hosted solution or connect to Jan via API for teams and enterprises. +We build in public. Everything from our model training to our product roadmap is open. - **[Learn more →](./platforms/server)** - - - -## What Makes Jan Different - -| Feature | Other AI Assistants | Jan | -| :--- | :--- | :--- | -| **Models** | Wrapper around Claude/GPT | Our own models + You can own them | -| **Dual mode** | Your data on their servers | Your data stays yours | -| **Deployment** | Cloud only | Local, self-hosted, or cloud | -| **Cost** | Subscription forever | Free locally, pay for cloud | - -## Development Timeline - -Jan is actively developed with regular releases. Our development follows these key milestones: - -### Current Focus -- **Jan Desktop**: Continuous improvements and model support -- **Jan Web**: Beta launch preparation -- **Model Development**: Jan Nano optimization and v1 launch - -### Next 6 Months -- Jan Web public beta -- Mobile app development -- Server deployment tools - -### Future Vision -- Complete full-stack AI solution -- Advanced tool integration -- Enterprise features - - +- [GitHub](link) - Contribute code +- [Handbook](link) - See how we train models +- [Discord](link) - Join the discussion diff --git a/website/src/content/products/models/jan-v1.mdx b/website/src/content/products/models/jan-v1.mdx index 8f17b07e4..bde5817b2 100644 --- a/website/src/content/products/models/jan-v1.mdx +++ b/website/src/content/products/models/jan-v1.mdx @@ -1,29 +1,51 @@ --- -title: Jan V1 -description: Our upcoming family of foundational models, built to compete with the best. +title: Jan Models +description: Specialized AI models trained in public for real tasks. sidebar: order: 1 banner: - content: 'In Development: Jan V1 models are currently being trained and are not yet available.' + content: 'Watch live training progress below. First models releasing Q1 2025.' --- import { Aside } from '@astrojs/starlight/components'; -## Our Foundational Model Family +## Not Just Another Model Family -Jan V1 is our in-house, still in training, family of models designed to compete directly with leading models. We're building powerful, general-purpose models from the ground up to solve real-world problems with a focus on efficiency and privacy. +Jan Models aren't general-purpose chatbots. Each model is trained for specific tasks that matter in daily work: search, analysis, creative writing, coding, research. They work together, each handling what it does best. -### Planned Model Lineup +### Current Training Status -| Model | Target Size | Intended Use Case | Availability | -|:------------|:------------|:-----------------------------|:--------------| -| Jan V1-7B | 4-8GB | Fast, efficient daily tasks | Coming Soon | -| Jan V1-13B | 8-16GB | Balanced power and performance | Coming Soon | -| Jan V1-70B | 40-64GB | Deep analysis, professional work | Coming Soon | -| Jan V1-2350B | 100GB+ | Frontier research, complex tasks | Planned 2026 | +| Model | Specialization | Size | Training Progress | Status | +|:------|:--------------|:-----|:------------------|:-------| +| Jan-Search | Web search + synthesis | 7B | ████████░░ 82% | Testing phase | +| Jan-Write | Creative + technical writing | 13B | ████░░░░░░ 41% | Active training | +| Jan-Analyze | Data analysis + reasoning | 13B | ██░░░░░░░░ 23% | Dataset prep | +| Jan-Code | Code generation + debugging | 7B | ░░░░░░░░░░ 0% | Starting Jan 2025 | -### What to Expect -- **Competitive Performance**: Aiming for results on par with leading closed-source models. -- **Optimized for Local Use**: Efficient quantized versions for running on your own hardware. -- **Privacy-Centric**: Trainable and runnable in your own environment, ensuring your data stays yours. -- **Seamless Integration**: Designed to work perfectly within the Jan ecosystem. -- **Fine-tuning support**: Easy to adapt to specific tasks or domains. \ No newline at end of file + + +### Why Specialized Models? + +One model can't excel at everything. GPT-4o, o3, o4, or Claude 4 Sonnet writing poetry +use the same weights to do math as well, which can be inefficient and expensive. + +Our approach: +- **Jan-Search** knows how to query, crawl, and synthesize +- **Jan-Write** understands tone, structure, and creativity +- **Jan-Analyze** excels at reasoning and data interpretation +- Models work together through the Jan orchestration layer + +### Built for the Ecosystem + +These aren't standalone models. They're designed to: +- Run efficiently on local hardware (quantized to 4-8GB) +- Work with Jan Tools (browser automation, file parsing, memory) +- Scale from laptop to server without code changes +- Share context and hand off tasks to each other + +### Help Us Improve + +Models are only as good as their real-world performance. [Test our models](link) against your actual use cases and vote on what works. + +We train on your feedback, not just benchmarks. diff --git a/website/src/content/products/platforms/desktop.mdx b/website/src/content/products/platforms/desktop.mdx index 9d0e65bd5..b77cf9c72 100644 --- a/website/src/content/products/platforms/desktop.mdx +++ b/website/src/content/products/platforms/desktop.mdx @@ -1,113 +1,111 @@ --- title: Jan Desktop -description: AI that runs on your computer, not someone else's. Your personal AI workstation. +description: The foundation of your AI ecosystem. Runs on your hardware. sidebar: order: 2 --- import { Aside, Card, CardGrid, Tabs, TabItem } from '@astrojs/starlight/components'; -This is how Jan started and it has been available since day 1. +Jan Desktop is your local AI workstation. Download it, run your own models, or connect to cloud providers. Your computer, your choice. -Jan Desktop strives to be: +## How It Works -> Your personal AI workstation that helps with your use cases and powers other devices. Run models locally right away or bring an API key to connect to your favorite cloud-based models. +### Default: Local Mode +Open Jan. Start chatting with Jan Nano. No internet, no account, no API keys. Your conversations never leave your machine. -Jan Desktop is where it all starts. Download it, open it, and start chatting. Your AI runs on your computer with zero setup required. +### Optional: Cloud Mode +Need more power? Connect to: +- Your own Jan Server +- jan.ai (coming soon) +- Any OpenAI-compatible API -## Two Modes, Zero Complexity - -### Local Mode (Default) -Your conversations stay on your computer. No internet needed. Complete privacy. - -### Cloud Mode -Connect to more powerful models when you need them. Your choice of provider. - -