Back to Tools
Goodfire vs Recursion
Side-by-side comparison of features, pricing, and ratings
Interpretability platform to understand, debug, and steer AI models with software-like precision.
Visit Website
Pricing
Contact Sales
Contact Sales
Plans
—
—
Popularity
6.7k views
7.0k views
Skill Level
Advanced
Advanced
API Available
Platforms
API
—
Categories
💻 Code & Development📊 Data & Analytics🔬 Research & Education
📊 Data & Analytics🏥 Healthcare
Features
Reverse engineer causal mechanisms in LLMs and vision models
Identify and remove confounders in model behavior
Detect performative chain-of-thought and enable early exit (68% token savings)
Use interpretability features as rewards to reduce hallucinations (58% reduction)
Self-correcting search for materials discovery via internal feedback
Analyze latent space of vision models (e.g., cardiac echo, genomics)
Debug information bottlenecks in robotics policies
Validate domain understanding (e.g., clinical knowledge in medical AI)
Steer model internals for targeted performance improvements
Harvest activations from trillion-parameter models for interpretability
Open-source sparse autoencoders for Llama 3.3 70B and Llama 3.1 8B
Feature steering for reliable and expressive AI engineering
AI-driven drug discovery using cellular imaging
Recursion Operating System (OS) platform
Automated wet lab with robotics and computer vision
50+ petabytes of proprietary biological and chemical data
Machine learning models for target identification
Molecule design optimization
Phenomics, transcriptomics, proteomics, ADME data
De-identified patient data integration
BioHive-2 supercomputer (built with NVIDIA)
Pipeline of potential first-in-class and best-in-class treatments
Phase 1/2 clinical trials for oncology and rare diseases
Partnerships with pharmaceutical and technology leaders
Continuous feedback loop for model improvement
Integrations
Llama 3.3 70B
Llama 3.1 8B
Evo 2
EchoJEPA
Radical AI
NVIDIA
Robotics and computer vision systems