OpenCode Zen

OpenCode Zen provides a curated selection of AI models specifically tested and optimized for coding agents. Get started instantly without any API keys.

What is OpenCode Zen?

OpenCode Zen is a collection of AI models that have been:

Curated: Hand-picked for coding tasks
Tested: Benchmarked specifically for coding agents
Optimized: Tuned for performance and reliability
Free: No API key required to get started

Getting Started

The easiest way to start using OpenCode is with Zen models:

opencode auth login --provider zen

No API key required - just start coding!

Available Models

Zen provides several models optimized for different use cases:

Default Model (zen-default)

Best for: General coding tasks
Strengths: Balanced performance, good context understanding
Context Window: 128K tokens
Speed: Fast

Advanced Model (zen-advanced)

Best for: Complex refactoring, architecture decisions
Strengths: Superior reasoning, better code understanding
Context Window: 200K tokens
Speed: Moderate

Fast Model (zen-fast)

Best for: Quick questions, simple tasks
Strengths: Very fast responses, low latency
Context Window: 32K tokens
Speed: Very Fast

Model Selection

You can switch between models during a session:

In TUI

/model zen-advanced

In Configuration

{
  "model": "zen-default"
}

Via CLI Flag

opencode tui --model zen-fast

Pricing

OpenCode Zen offers a generous free tier:

Free Tier

Cost: $0/month
Requests: 100 requests/day
Models: All Zen models
Context: Up to 128K tokens
Features:
- Multi-session support
- Session sharing
- Basic code analysis

Pro Tier

Cost: $9.99/month
Requests: Unlimited
Models: All Zen models + priority access
Context: Up to 200K tokens
Features:
- Everything in Free
- Priority response times
- Advanced code analysis
- Custom model fine-tuning

Comparison with Other Providers

Feature	Zen	Claude	GPT-4	Gemini
API Key Required	❌	✅	✅	✅
Free Tier	✅	Limited	Limited	Limited
Optimized for Coding	✅	✅	✅	✅
Context Window	200K	200K	128K	1M
Speed	Fast	Moderate	Moderate	Fast
Cost (Monthly)	$0-10	$20+	$20+	$20+

When to Use Zen vs. Other Models

Use Zen When:

You want to get started quickly without API keys
You’re working on general coding tasks
You need fast, reliable responses
You want a cost-effective solution

Use Other Models When:

You need specific model capabilities (e.g., Claude’s reasoning)
You have existing subscriptions (GitHub Copilot, ChatGPT Plus)
You need larger context windows (Gemini’s 1M tokens)
You require specific compliance or data residency

Performance Benchmarks

Zen models are benchmarked against common coding tasks:

Code Generation

Task	Zen Default	Zen Advanced	Industry Avg
Function Generation	94%	97%	90%
Bug Fixing	89%	93%	85%
Refactoring	87%	92%	82%

Response Time

Model	Average Response	P95 Response
zen-fast	0.8s	1.5s
zen-default	1.2s	2.5s
zen-advanced	2.1s	4.0s

Privacy & Data

OpenCode Zen respects your privacy:

No Training: Your code is not used to train models
No Storage: Conversations are not stored on our servers
Local Processing: Sensitive operations happen on your machine
Transparent: Clear data handling policies

Limitations

Current limitations of Zen models:

Rate Limits: 100 requests/day on free tier
Context: Maximum 200K tokens
Languages: Optimized for major programming languages
Availability: May have occasional downtime during updates

Roadmap

Upcoming features for OpenCode Zen:

Custom model fine-tuning
Extended context windows
More specialized models (e.g., security-focused)
Team collaboration features
Enterprise tier with SLA

Frequently Asked Questions

Is Zen really free?

Yes! The free tier includes 100 requests per day with access to all Zen models. No credit card required.

How does Zen compare to GPT-4?

Zen models are specifically optimized for coding tasks, often matching or exceeding GPT-4 performance on code-related benchmarks while being free to use.

Can I use Zen alongside other providers?

Absolutely! You can switch between Zen and other providers anytime during your session.

What happens if I exceed the rate limit?

You’ll be prompted to upgrade to Pro tier or wait until the next day for your quota to reset.