Skip to main content
AINative Studio
Products▾
Solutions▾
AI for BusinessNewFor DevelopersPricingDocs
Sign InBook a Call

AI Coding Assistant
for the Terminal

19 Coding Models. Zero Config. Persistent Memory.

Code faster with 19 coding models, persistent memory, and Agent Cloud integration. The AI coding assistant that actually remembers your stack.

DocumentationContact Sales
16,000+ developersFree CLI to install3-day free trial, then $5/mo

Used by developers from leading tech companies and universities

Get started in seconds

npmnpx @ainative/cody-cli
Homebrewbrew tap AINative-Studio/tap && brew install AINative-Studio/tap/cody
curlcurl -fsSL https://ainative.studio/install-cody.sh | sh

Everything you need, nothing you don't

A complete AI engineering environment that runs where you already work.

19 Coding Models

From Qwen3 Coder 30B at $0.04/1M to flagship GLM-5, Kimi K3 and DeepSeek V4 Pro. Switch models mid-session with /config model=.

Built-in Tools

Read, Write, Edit, Bash, Glob, Grep — no plugins needed. Full filesystem access out of the box.

MCP Support

Connect to any MCP server for extended capabilities. Database, memory, APIs — all composable.

TDD Enforced

Built-in skills for test-driven development. Write tests first, then implement — enforced by default.

Multi-Platform

macOS, Linux, native binaries, or via npm. Works wherever you have a terminal.

Secure by Design

Sandboxed execution, permission-based tool access, and full audit logs for every action taken.

New in v0.11

Latest Features

Autonomous goals, background agents, security scanning, and more.

Smart Permissions

/auto on

Safe actions run without prompts — dangerous ones still ask.

Autonomous Goal Mode

/goal <condition>

Cody works autonomously until your condition is met.

Agent Monitor

/agent-view

See all running, completed, and failed agents in one view.

YAML Pipelines

/workflow run

Define multi-step agent pipelines in declarative YAML.

PR Review & Auto-Fix

/pr-review + /pr-fix

Automated pull request review with one-click fixes.

Security Scanning

/security-review

OWASP Top 10 vulnerability scanning built into your workflow.

Background Agents

/bg <task>

Kick off tasks in the background — keep coding while they run.

Code Review

/code-review

Review staged changes for bugs with precise line citations.

What makes Cody different

Cody is built on AINative infrastructure — not a thin wrapper on top of a third-party API.

ZeroDB Persistent Memory

Exclusive

Cody remembers across sessions. Every decision, every pattern, every codebase convention — stored in ZeroDB and recalled automatically. No more re-explaining your stack every session.

ZeroMemory docs
  • Multi-tier cognitive memory: working, episodic, semantic
  • Automatic memory consolidation and decay scoring
  • GraphRAG hybrid search across your entire history
  • Shared memory across team members on Business/Enterprise

Agent Cloud Integration

Enterprise

Promote Cody tasks to cloud-scale. Kick off long-running jobs from your terminal, monitor them from anywhere, and let Agent Cloud handle parallelization and retry logic.

Agent Cloud docs
  • Deploy terminal tasks to managed agent infrastructure
  • A2A networking: coordinate Cody with other agents
  • Auto-scaling — run 100 tasks in parallel
  • Full observability: traces, logs, and cost breakdowns

See it in action

~/my-project
$ cody -p "Write a Python function to check if a number is prime"
Cody > I'll create a prime checker for you.
def is_prime(n: int) -> bool:
    if n <= 1:
        return False
    if n <= 3:
        return True
    if n % 2 == 0 or n % 3 == 0:
        return False
    i = 5
    while i * i <= n:
        if n % i == 0 or n % (i + 2) == 0:
            return False
        i += 6
    return True
Wrote to prime.py · Stored pattern in ZeroDB memory
$ █

19 coding models, zero config

Switch models with a single flag. All prices per 1M tokens (verified 2026-07-28), all hosted on AINative infrastructure — no third-party API keys needed. Coding inference from $0.04 / 1M tokens, 5 models on the $5/mo Hobbyist plan (3-day free trial).

ModelPlanInput / 1MOutput / 1MContext
qwen3-coder-30b(cheapest)Hobbyist$0.04$0.04256K
qwen3-coder-next-80bHobbyist$0.06$0.06256K
gpt-oss-120bHobbyist$0.42$0.9064K
gemma-4-31bHobbyist$0.99$1.4964K
glm-4.7-previewHobbyist$1.25$2.758K
qwen-coder-7bPro$0.24$0.96128K
qwen3-coder-flashPro$0.15$0.60128K
deepseek-4-flashPro$0.20$0.80128K
kimi-k2Pro$0.25$1.00128K
kimi-k2.5Pro$0.25$1.00128K
kimi-k2.6Pro$0.25$1.00128K
kimi-k2-thinkingPro$0.25$1.00256K
qwen-coder-32bPro$0.28$1.10128K
minimax-m2.5Pro$0.30$1.20128K
minimax-m2.7Pro$0.60$2.40128K
qwen3.5-397bPro$0.30$1.20128K
deepseek-v4-proPro$0.44$0.87128K
glm-5Pro$0.60$2.20128K
glm-5.1Pro$0.60$2.20128K
glm-5.2Pro$0.60$2.20128K
kimi-k3Pro$0.60$2.50256K
claude-sonnet-4.5Pro$6.00$30.00195K
claude-opus-4Enterprise$15.00$75.00195K

All models available on Business and Enterprise plans. Model configuration docs

Frequently Asked Questions

Everything you need to know about Cody, AI IDEs, and how we compare to Cursor, Copilot, and more.

What is an AI IDE?
An AI IDE (Artificial Intelligence Integrated Development Environment) is a code editor or development environment that embeds AI models directly into the coding workflow. Unlike traditional IDEs, an AI IDE can read your codebase, write and edit files, run tests, and suggest architecture decisions autonomously. Cody is a terminal-native AI IDE with 19 coding models and persistent ZeroDB memory.
Is Cody a Cursor alternative?
Yes. Cody is a Cursor alternative built for developers who prefer the terminal. While Cursor is GUI-based, Cody runs natively in your terminal with full filesystem access, MCP server support, and persistent agent memory. The Cody CLI is free and open source to install, and you can start with a 3-day free trial of the $5/mo Hobbyist plan — making it one of the most affordable Cursor alternatives available.
What is the best affordable AI coding assistant?
Cody by AINative is one of the most affordable AI coding assistants available. The CLI is free and open source to install, and the $5/mo Hobbyist plan — which starts with a 3-day free trial — gives you access to multiple code and reasoning models, MCP support, TDD enforcement, and ZeroDB persistent memory.
How does Cody compare to GitHub Copilot?
Cody differs from GitHub Copilot in several key ways: it runs in your terminal rather than a GUI editor, supports 19 coding models, includes persistent memory that remembers your codebase across sessions, enforces test-driven development by default, and integrates with MCP servers. The Cody CLI is free and open source to install, and you can get started with a 3-day free trial of the $5/mo Hobbyist plan.
Can I use Cody offline?
Cody supports local model configurations for offline use. You can configure Cody to route requests to a locally-running model server. Cloud-based models and ZeroDB memory sync require internet connectivity, but core code editing workflows can run fully local.
What programming languages does Cody support?
Cody supports all major programming languages including Python, TypeScript, JavaScript, Go, Rust, Java, C++, Ruby, PHP, and more. Since Cody uses frontier coding models (GLM-5, Kimi K3, DeepSeek V4, Qwen3 Coder), it handles any language those models support. The built-in tools (Read, Write, Edit, Bash, Glob, Grep) work with any file type.

Start building with Cody

One command to install. Your code context remembered forever.

npx @ainative/cody-cli
DocumentationContact Sales
Built by AINative Studio