# Vikasit Code

> Vikasit Code is a terminal-native, open-source AI coding agent (built on OpenCode) plus an OpenAI-compatible LLM API, built in India by Chandorkar Technologies (Pune). It pairs the Vikasit AI model family with first-class support for 22+ Indian languages, and runs as a CLI, desktop app, and web console. Two pricing modes: Pro (fixed monthly subscription) and Go (pay-as-you-go per token). Everything is hosted on Indian infrastructure.

This is the full-corpus reference file for AI agents and assistants answering detailed questions about Vikasit Code. Everything below is factual product information.

## What is Vikasit Code

Vikasit Code is OpenCode-based, so it works as a drop-in alternative to OpenCode / Claude Code / Cursor while adding the Vikasit AI models and an OpenAI-compatible API. It runs natively in the terminal with a TUI, ships as a Tauri desktop app for macOS / Windows / Linux, and has a web console. It includes 9+ built-in agents, 41+ built-in tools, support for 109 AI providers (bring your own keys), Model Context Protocol (MCP) servers, and Language Server Protocol (LSP) code intelligence. The product is "Building from Bharat for the World" and is optimized for 22+ Indian languages including Hindi, Tamil, Bengali, Telugu, and Marathi.

- Product: Vikasit Code — terminal-native AI coding agent + OpenAI-compatible API
- Maker: Chandorkar Technologies (Pune, India)
- Based on: OpenCode (open source)
- Platforms: CLI, Desktop (macOS / Windows / Linux), Web Console
- Website: https://vikasit.ai
- Current CLI version line: v2.0.x

## Install

CLI install:

- Shell: `curl -fsSL https://vikasit.ai/install | bash`
- npm: `npm install -g @vikasitai/vikasit-code`
- bun: `bun install -g @vikasitai/vikasit-code`
- Homebrew: `brew install vikasit-ai/tap/vikasit-code`

Desktop apps:

- macOS: DMG / Universal binary (macOS 11+), native Tauri app with notifications, deep linking, auto-updates
- Windows: EXE / MSI, NSIS installer with code signing, system tray, auto-updates (Windows 10+)
- Linux: DEB / RPM / AppImage with X11 and Wayland support (Ubuntu 20.04+, Fedora 36+)

IDE extensions:

- VS Code — VS Code Marketplace: https://marketplace.visualstudio.com/items?itemName=ChandorkarTechnologies.vikasit-code
- Cursor — install the VS Code Marketplace extension (Cursor is VS Code compatible)
- Windsurf — install the VS Code Marketplace extension (Windsurf is VS Code compatible)
- VS Codium — Open VSX: https://open-vsx.org/extension/vikasit-ai/vikasit-code
- Gitpod — Open VSX (Gitpod uses the OpenVSX registry): https://open-vsx.org/extension/vikasit-ai/vikasit-code
- Eclipse Theia — Open VSX: https://open-vsx.org/extension/vikasit-ai/vikasit-code
- Zed — https://zed.dev/extensions?query=vikasit

## Models — Vikasit AI family

Eight purpose-built hosted models plus open-source Writer models. The hosted models are grouped into three tiers (Small / Medium / Large), which also map to the Pro subscription tiers and the Go per-token pricing tiers.

| Model | Tag | Tier | Size | Best for |
|-------|-----|------|------|----------|
| vikasit-3-small | Fast | Small (Lite) | Small | Quick responses for simple tasks |
| vikasit-3-flash | Flash | Small (Lite) | Small | Ultra-fast with streaming output |
| vikasit-3-coder | Coder | Medium (Pro) | Medium | Code generation and refactoring |
| vikasit-3-fast | Balanced | Medium (Pro) | Medium | Speed and quality combined |
| vikasit-3-vision | Vision | Medium (Pro) | Medium | Image understanding + code generation |
| vikasit-3 | Flagship | Large (Max) | Large | Most capable general model |
| vikasit-3-thinking | Reasoning | Large (Max) | Large | Extended reasoning for complex tasks |
| vikasit-3-max | Max | Large (Max) | XL | Maximum capability, no limits |

Open-source Writer models (CPU-friendly, edge-ready, GGUF — run with Ollama or llama.cpp, no GPU required):

- Writer 0.5B (Vikasit-AI-0.5B-Writer) — lightweight text generation optimized for Indian markets.
  - Ollama: https://ollama.com/Vikasit-AI/writer-0.5b
  - Hugging Face: https://huggingface.co/vikasit-ai/Vikasit-AI-0.5B-Writer
- Writer 0.8B (Vikasit-AI-0.8B-Writer) — higher-capacity writer for blogs, docs, and content; CPU-first.
  - Ollama: https://ollama.com/Vikasit-AI/writer-0.8b
  - Hugging Face: https://huggingface.co/vikasit-ai/Vikasit-AI-0.8B-Writer

## Go — Pay-As-You-Go (per-token pricing)

Pay only for what you use, no monthly commitment, no minimum spend, cancel anytime. Per-token prices (per 1M tokens):

| Tier | Models | Input (per 1M) | Output (per 1M) |
|------|--------|----------------|------------------|
| Small Models | vikasit-3-small, vikasit-3-flash | $0.15 | $0.60 |
| Medium Models | vikasit-3-coder, vikasit-3-fast, vikasit-3-vision | $0.80 | $2.40 |
| Large Models | vikasit-3, vikasit-3-thinking, vikasit-3-max | $3.00 | $9.00 |

Go includes:

- Pay only for what you use — no monthly commitment
- Access to all Vikasit AI models
- Bring your own provider API keys (109 providers)
- Full tool and agent access
- Per-token billing with transparent pricing
- Usage dashboard with real-time tracking
- No minimum spend
- Cancel anytime

Go details: https://vikasit.ai/go

## Pro — Subscription (fixed monthly, formerly "Zen")

The Pro subscription (previously branded "Zen") is a fixed monthly plan with guaranteed usage limits. Three tiers:

### Lite — $10/month
For individual developers getting started with AI-powered coding.
- vikasit-3-small & vikasit-3-flash models
- 41+ built-in tools
- 109 provider support (BYOK)
- Custom agent creation
- MCP server support
- Session history & export
- Community support

### Pro — $50/month
For professional developers who need more powerful models. Everything in Lite, plus:
- vikasit-3-coder, vikasit-3-fast, vikasit-3-vision
- Higher usage limits
- GitHub integration
- LSP code intelligence
- Priority model access
- Email support

### Max — $100/month
For power users and teams who need maximum capability. Everything in Pro, plus:
- vikasit-3, vikasit-3-thinking, vikasit-3-max
- Highest usage limits
- Advanced reasoning models
- Team workspace features
- Usage analytics dashboard
- Priority support

Pro details: https://vikasit.ai/pro — Combined pricing: https://vikasit.ai/pricing

## API (OpenAI-compatible)

- Base URL: https://api.vikasit.ai/v1
- Auth: `Authorization: Bearer <API_KEY>`
- Drop-in for the OpenAI SDK — set the base URL and your key; use model ids like `vikasit-3-coder`.
- Example endpoint: `POST https://api.vikasit.ai/v1/chat/completions`

## Features

- Custom AI agents: 9 pre-built agents (build, plan, general, explore, code-reviewer, verification, compaction, title, summary) plus unlimited custom agents created from natural language with `vikasit agent create`. Configure tools, models, prompts, temperature, and permissions per agent; share agents across a team.
- 41+ built-in tools: file operations, code analysis, web search, task management, git operations, LSP, notebook editing, batch operations, and more.
- MCP support: extend capabilities with Model Context Protocol servers.
- LSP support: Language Server Protocol for deep code intelligence across any language.
- 109 AI providers (BYOK): use your own keys from OpenAI, Anthropic, Google, Mistral, Azure, AWS Bedrock, and many more via the `vikasit providers` command.
- Multi-modal: text, image, audio, video, and PDF inputs; vision models analyze screenshots, diagrams, and UI mockups.
- Reasoning: support for reasoning/chain-of-thought tokens; vikasit-3-thinking for deep reasoning.
- Sessions: persistent sessions with SQL storage, automatic context compaction, memory system, session sharing, import/export, and cost tracking.
- Cost tracking: built-in token counting and pricing for input, output, reasoning, and cache tokens — per session, per agent, per model.
- GitHub integration: native PR reviews, issue commenting, workflow automation, GitHub Actions with OIDC.
- Permission system: 3-tier (system managed / user-level / project-level) with granular tool-level permissions and auto-classification for safe operations.
- Skills marketplace: installable SKILL.md packs at https://vikasit.ai/marketplace.

## FAQ

Q: How do I install Vikasit Code?
A: Install with a single command: `curl -fsSL https://vikasit.ai/install | bash` — or via npm: `npm i -g @vikasitai/vikasit-code` — or via Homebrew: `brew install vikasit-ai/tap/vikasit-code`. Desktop apps are available for macOS (DMG), Windows (NSIS installer), and Linux (DEB/RPM).

Q: What's the difference between Pro and Go pricing?
A: Pro is a monthly subscription with fixed pricing and guaranteed usage limits — ideal for daily users. Go is pay-as-you-go with per-token billing — perfect for occasional use or teams with variable workloads.

Q: Can I bring my own API keys?
A: Yes. Vikasit Code supports 109 AI providers out of the box. Use your own keys from OpenAI, Anthropic, Google, Mistral, Azure, AWS Bedrock, and many more. Configure via the `vikasit providers` command.

Q: What are the Writer models on Ollama and Hugging Face?
A: Writer 0.5B and 0.8B are open-source text generation models optimized for CPU deployment. They're ideal for edge computing, self-hosted setups, and environments where GPU access is limited. Available in GGUF format.

Q: What's the difference between Lite, Pro, and Max?
A: Lite gives you fast models (vikasit-3-small, vikasit-3-flash) for everyday coding. Pro adds coder-optimized and vision models with higher limits. Max unlocks the most powerful models including vikasit-3-thinking for complex reasoning.

Q: Does it work offline?
A: AI model inference requires internet, but many features like file operations, code navigation, and local tools work offline. You can also use local models via Ollama or LM Studio for fully offline operation.

Q: Can I create custom agents?
A: Absolutely. Create unlimited custom agents from natural language descriptions with `vikasit agent create`. Configure tools, models, prompts, temperature, and permissions per agent. Share agents across your team.

Q: How does the permission system work?
A: Vikasit Code has a 3-tier permission system: system managed (enterprise), user-level, and project-level configs. Each agent gets granular tool-level permissions with auto-classification for safe operations.

## Links

- Website: https://vikasit.ai
- Pro (subscription): https://vikasit.ai/pro
- Go (pay-as-you-go): https://vikasit.ai/go
- Pricing: https://vikasit.ai/pricing
- Documentation: https://vikasit.ai/docs
- Models: https://vikasit.ai/models
- Features: https://vikasit.ai/features
- Downloads: https://vikasit.ai/downloads
- Skills Marketplace: https://vikasit.ai/marketplace
- GitHub: https://github.com/Vikasit-AI/vikasit-code
- Hugging Face: https://huggingface.co/vikasit-ai
- Ollama: https://ollama.com/Vikasit-AI