For the first time, cannabis operators no longer have to choose between handcrafted experience and disconnected academic theory. Newton Insights today released the most advanced cannabis-specific AI model ever developed — a domain-trained LLM built on 400+ peer-reviewed studies that generated 160,000 expert-level Q&As. The result: instant, science-backed answers to the toughest extraction and formulation questions, all tailored to the realities of real-world operations.
"Cannabis teams are drowning in data but starving for clarity," said Bryan Fields, Co-Founder of Newton Insights. "The industry has been stuck choosing between academic theory and trial-and-error experience."
"This LLM changes that equation completely," adds Co-Founder Kellan Finney. "We're giving extraction teams the scientific foundation they need to optimize their processes without the guesswork."
Key Highlights & Model Comparisons
- 161,170 expert-level Q&As covering cannabis chemistry, biology, and extraction
- 83.6% accuracy on scientific benchmarks — outperforming GPT-4 (60.2%) and Claude 3 (66.7%) on cannabis-specific tasks
- Trained for depth, not breadth — a domain expert purpose-built for cannabis operations
- Built from 400+ peer-reviewed studies and 127M+ science tokens
- Outperforms other 8B-class models in science Q&A by 10–20 points — despite a smaller size and tighter domain scope
Real-World Applications for Extraction Teams
- Optimize extraction parameters using validated scientific inputs
- Reduce waste by minimizing trial-and-error and dialing in system performance
- Proactively guide strain-to-yield decisions with AI-powered analysis and literature-backed insights
What’s Next
This LLM is the foundation of Newton's full-stack Vertical Software AI platform. Future releases will integrate real-time process analytical data (PAT) and a physics-based Digital Twin to create a fully autonomous feedback loop for cannabis operations.About Newton Insights
Newton Insights builds AI-powered tools for the cannabis industry, starting with extraction. Our mission is to help teams move from reactive guesswork to proactive control — grounded in validated science and world-class technology.
"This LLM changes that equation completely, We're giving extraction teams the scientific foundation they need to optimize their processes without the guesswork."- Co-Founder Kellan Finney
Contact:
Bryan Fields
Newton@8threv.com
