Foundation Model Transparency Index
by Stanford CRFMPublished 2025-12-01Source ↗Overall:95
- Agent Protocols
- 100
- AI bug bounty
- 100
- Amount of usage
- 100
- AUP enforcement frequency
- 0
- AUP enforcement process
- 100
- Basic model properties
- 100
- Benchmarked inference
- 100
- Benefits Assessment
- 100
- Capabilities evaluation
- 100
- Capabilities taxonomy
- 100
- Carbon emissions for final training run
- 100
- Change log
- 100
- Classification of usage data
- 100
- Code access
- 100
- Compute hardware for final training run
- 100
- Compute provider
- 100
- Compute usage for final training run
- 100
- Compute usage including R&D
- 100
- Consumer/enterprise usage
- 100
- Crawling
- 100
- Data acquisition methods
- 100
- Data domain composition
- 100
- Data laborer practices
- 100
- Data language composition
- 100
- Data processing methods
- 100
- Data processing purpose
- 100
- Data processing techniques
- 100
- Data replicability
- 100
- Data retention and deletion policy
- 100
- Data size
- 100
- Deeper model properties
- 100
- Detection of machine-generated content
- 100
- Development duration for final training run
- 100
- Distribution channels with usage data
- 100
- Documentation for responsible use
- 100
- Downstream
- 91.7
- Acceptable use policy
- 80
- Accountability
- 100
- Downstream mitigations
- 100
- Impact
- 85.7
- Model Behavior Policy
- 100
- Post-deployment monitoring
- 100
- Usage data
- 80
- Energy usage for final training run
- 100
- Enterprise mitigations
- 100
- Enterprise users
- 0
- External data access
- 100
- External developer mitigations
- 100
- External products and services
- 100
- External reproducibility of capabilities evaluation
- 100
- External reproducibility of mitigations evaluation
- 0
- External reproducibility of risks evaluation
- 100
- External risk evaluation
- 100
- Feedback mechanisms
- 100
- Foundation model roadmap
- 100
- Geographic statistics
- 0
- Government commitments
- 100
- Government use
- 100
- Instructions for data generation
- 100
- Intermediate tokens
- 100
- Internal compute allocation
- 100
- Internal product and service mitigations
- 100
- Internal products and services
- 100
- Licensed data compensation
- 100
- Licensed data sources
- 100
- Misuse incident reporting protocol
- 100
- Mitigations efficacy
- 100
- Mitigations taxonomy
- 100
- Mitigations taxonomy mapped to risk taxonomy
- 100
- Model
- 93.3
- Capabilities
- 75
- Model cost
- 100
- Model dependencies
- 100
- Model access
- 100
- Model information
- 100
- Model Mitigations
- 80
- Model objectives
- 100
- Release
- 100
- Model response characteristics
- 100
- Risks
- 100
- Model stages
- 100
- Model theft prevention measures
- 100
- New human-generated data sources
- 100
- Notice of usage data used in training
- 100
- Open weights
- 100
- Organization chart
- 100
- Oversight mechanism
- 100
- Permitted and prohibited users
- 100
- Permitted, restricted, and prohibited model behaviors
- 100
- Permitted, restricted, and prohibited uses
- 100
- Post-deployment coordination with government
- 100
- Pre-deployment risk evaluation
- 100
- Public datasets
- 100
- Quantization
- 100
- Regional policy variations
- 100
- Release stages
- 100
- Researcher credits
- 100
- Responsible disclosure policy
- 100
- Risks evaluation
- 100
- Risks taxonomy
- 100
- Risk thresholds
- 100
- Safe harbor
- 100
- Security incident reporting protocol
- 100
- Specialized access
- 100
- Synthetic data purpose
- 100
- Synthetic data sources
- 100
- System prompt
- 100
- Terms of use
- 100
- Top distribution channels
- 100
- Train-test overlap
- 0
- Upstream
- 100
- Compute
- 100
- Data Acquisition
- 100
- Data Processing
- 100
- Data Properties
- 100
- Methods
- 100
- Other resources
- 100
- Usage data used in training
- 100
- Users of internal products and services
- 100
- Versioning protocol
- 100
- Water usage for final training run
- 100
- Whistleblower protection
- 100
Grade trajectory (2 waves)
- v1.2 December 2025latest
- 95
- v1.1 May 2024
- 126.7