Obsidian-32B-Instruct
Obsidian-32B-Instruct is our most advanced reasoning model, engineered for complex problem-solving across mathematics, coding, scientific analysis, and real-world task execution. Built on a foundation of 32.8 billion parameters with a 128,000-token context window, Obsidian delivers industry-leading performance on challenging benchmarks while maintaining exceptional efficiency. Its hybrid thinking architecture seamlessly switches between deep analytical reasoning for complex problems and rapid inference for straightforward tasks, making it the ideal choice for enterprise applications requiring both accuracy and speed.
Deep Reasoning Performance
Comparison across complex reasoning benchmarks
Obsidian is available on TorinChat, our API, and our major cloud platform, DreamDev. For developers, use via the Dominion Labs API. Pricing is now $5/$15 per million tokens—making Obsidian accessible to even more users, teams, and enterprises.
Alongside Obsidian, we're actively updating the Dominion Labs Developer Platform, DreamDev, and TorinChat. See our product-focused section below for details.
What Makes Obsidian Exceptional
Obsidian represents a breakthrough in AI reasoning, delivering enterprise-grade performance across the most demanding tasks while maintaining exceptional efficiency.
Hybrid Thinking Modes
Seamlessly switch between deep reasoning for complex tasks and fast responses for everyday queries—all within a single model.
Enterprise-Grade Performance
Industry-leading benchmarks across coding, mathematics, and reasoning tasks. Outperforms larger models on critical evaluations.
Multilingual Excellence
Native support for 119 languages with consistent performance across diverse linguistic contexts and technical domains.
Industry-Leading Performance
Obsidian achieves state-of-the-art results across industry-standard benchmarks, demonstrating superior reasoning, coding ability, and problem-solving capabilities.
Switch Between Deep Reasoning and Fast Responses
Obsidian's unique hybrid architecture allows you to choose the right mode for your task—eliminating the need for separate models.
Thinking Mode
Activates chain-of-thought reasoning for complex challenges requiring deep analysis, multi-step problem-solving, and careful consideration.
- Mathematical proofs and derivations
- Complex code architecture decisions
- Multi-step logical reasoning
- Research and analysis tasks
Fast Mode
Optimized for efficient, concise responses to straightforward queries and general-purpose tasks requiring quick turnaround.
- General knowledge questions
- Code completion and suggestions
- Document summarization
- Conversational interactions
Built for Complex Challenges
Obsidian delivers exceptional performance across diverse capabilities, from advanced coding to multilingual communication. See how it compares to leading models.
| Capability | Obsidian | Qwen3 32B | Claude Opus 4.5 | GPT-4o | DeepSeek-V3 |
|---|---|---|---|---|---|
| Code generation and debugging | 89% | 87% | 92% | 88% | 91% |
| Mathematical reasoning | 91% | 94% | 90% | 87% | 88% |
| Multilingual communication | 94% | 95% | 89% | 90% | 85% |
| Data analysis | 87% | 89% | 91% | 90% | 86% |
| Technical documentation | 90% | 88% | 93% | 89% | 87% |
| Research and planning | 88% | 90% | 92% | 88% | 85% |
| Novel problem solving | 92% | 93% | 91% | 86% | 87% |
| Visual reasoning | — | 88% | 94% | 93% | — |
| Tool use | 85% | 86% | 90% | 91% | 82% |
Scores based on internal benchmarks and third-party evaluations. "—" indicates capability not yet available.
Technical Specifications
Seamless Integration
Access Obsidian through our TorinAI platform or integrate directly via API. Built for production with enterprise-grade reliability.
Key Features
- REST API — Simple HTTP endpoints for easy integration
- Streaming Support — Real-time token streaming for responsive UX
- Function Calling — Native tool use and structured outputs
- Low Latency — Optimized infrastructure for fast response times
- Enterprise SLAs — 99.9% uptime guarantee
Quick Start Example
// Initialize Obsidian
const response = await fetch('https://api.dmnlabs.org/v1/chat', {
method: 'POST',
headers: {
'Content-Type': 'application/json',
'Authorization': 'Bearer YOUR_API_KEY'
},
body: JSON.stringify({
model: 'obsidian-32b-instruct',
messages: [{
role: 'user',
content: 'Explain quantum entanglement'
}],
thinking_mode: 'deep' // or 'fast'
})
});
const data = await response.json();
console.log(data.response);Experience Obsidian Today
Try Obsidian in TorinChat or integrate it into your applications via our API.
Free tier available. No credit card required to start.