Knowledge Library
Every attribute sourc.dev tracks has an explanation here. Definitions built for developers, not marketing teams.
LLM Metrics
Context window
Maximum tokens the model can hold in context — input + output combined
tokens read →
Max output tokens
Maximum number of tokens the model can generate in a single response
tokens read →
Input price (per 1M tokens)
Cost in USD to send 1 million input tokens
USD read →
Output price (per 1M tokens)
Cost in USD to receive 1 million output tokens
USD read →
Speed (tokens/sec)
Average output generation speed in tokens per second
tps read →
MMLU benchmark
Massive Multitask Language Understanding — 57-subject knowledge test, 0-100%
% read →
HumanEval benchmark
OpenAI code generation benchmark — pass@1 on 164 Python problems, 0-100%
% read →
MATH-500
Competition maths % correct
% read →
GPQA Diamond benchmark
Graduate-level expert questions across biology, chemistry, physics
% read →
MATH benchmark
Competition mathematics benchmark
% read →
Jan 2023 Baseline Price
USD/1M tokens read →
Drift & Intelligence
Drift Index
Behavioural consistency score 0-100. Lower = more drift from baseline.
read →
Canary 7d
7-day canary prompt consistency — % of canary prompts returning expected output
% read →
Canary 30d
30-day canary consistency — % of canary prompts returning expected output
% read →
Silent updates (90d)
Count of unannounced behaviour changes detected by canary pipeline in 90 days
count read →
Self-knowledge accuracy
Accuracy when asked questions about its own documented specification
% read →
Prompt Intelligence Index
Composite real-world prompt handling score 0-100 (sourc.dev proprietary)
read →
Ambiguity handling
Score 0-100: handles unclear or incomplete prompts well
read →
Assumption transparency
Score 0-100: explicitly states assumptions made
read →
Clarification calibration
Score 0-100: asks for clarification when needed and only when needed
read →
Prompt repair score
Score 0-100: maintains quality on malformed or degraded prompts
read →
Compliance & Access
EU data residency
Data stored within EU borders by default without requiring upgrade
read →
Free tier
Has a usable free tier with meaningful limits
read →
MCP support
Supports Anthropic Model Context Protocol
read →
Open source
Source code publicly available (MIT, Apache, or similar)
read →
Open weights
Model weights publicly available for download and self-hosting
read →
API available
Can be accessed via a documented public API
read →
Function calling
Supports structured tool/function calling
read →
Streaming
Token streaming via SSE
read →
Vision / image input
Accepts image input
read →
SCIM Provisioning
read →
Performance
Lighthouse PWA score
Google Lighthouse PWA audit score 0-100
read →
30d uptime
Measured uptime percentage over the last 30 days
% read →
90d uptime
Measured uptime percentage over the last 90 days
% read →
Lighthouse Performance
Google Lighthouse performance score 0-100
read →
Lighthouse Accessibility
Google Lighthouse accessibility score 0-100
read →
Response time (ms)
Average API response time in milliseconds
ms read →
sourc.dev Metrics
Ecosystem Gravity
How strongly the ecosystem gravitates toward this entity — integrations, tool support, developer adoption
read →
Operator Trust Score
Business reliability composite: SLA, uptime history, pricing stability, documentation quality
read →
Developer Friction Index
Composite ease-of-start score: time to first API call, documentation clarity, SDK quality
read →
Momentum Score
Rate of positive change: integrations added, price reductions, feature releases, community growth
read →
EU Readiness Score
Composite EU enterprise deployment score: data residency, GDPR docs, AI Act compliance posture
read →
Integrations count
Number of known direct integrations with other tools and services
count read →
Certified integrations
Official integrations acknowledged by provider
count read →
Community integrations
Community-built integrations (unofficial)
count read →
General
Product status
Current lifecycle status of the model or product
read →
Target audience
Primary intended user group
read →
Region
Geographic scope of the data or service
read →
Pricing tier
Overall pricing model classification
read →
Parameters
Model size in billions of parameters
B read →
Auth method
Authentication method required to access API
read →
Provider
Company or organisation that develops and maintains this entity
read →
Model family
Model generation or family name (e.g. GPT-4 family, Claude 3 family)
read →
Release date
Date the model or product was publicly available
read →
Licence
SPDX identifier string
read →
Cache read price
USD per 1M cached input tokens
USD read →
Cache write price
USD per 1M cache write tokens
USD read →
E2E response time
Seconds to receive 500 output tokens
s read →
Avg reasoning tokens
Mean thinking tokens before answer
tok read →
Zero Data Retention
No prompt storage committed
read →
SWE-bench
Software engineering tasks %
% read →
AIME
Olympiad maths % correct
% read →
Chat arena ELO
LMArena text leaderboard score
read →
Code arena ELO
LMArena code leaderboard score
read →
AA Intelligence Index
Artificial Analysis composite score
read →
sourc Value Index
read →
Normalized SVI
read →
Price Deflation Index
read →
Value Density Score
read →
Integration Density Score
read →
Provider Concentration Risk
read →
Openness score
sourc.dev composite 0-90 (9 dimensions)
read →
LMSYS ELO
read →
Full Access Price (USD/month)
USD/month read →
Pricing Tier Tracked
read →
Price Verified Date
read →
Custom Pricing Only
read →
Pricing model
read →
Primary use case
read →
Rate limit (req/min)
rpm read →
Blast radius
read →
MCP servers published
Number of MCP servers this entity makes available to others. High count = infrastructure status.
count read →
MCP servers connected
Number of MCP servers this entity connects to as a client. High count = platform breadth.
count read →
On MCP directory
Listed on Anthropic official MCP server directory at modelcontextprotocol.io/servers
read →
MCP version
MCP specification version supported
read →
Founded
read →
Total funding (USD)
USD read →
Team size
read →
MCP servers
Official MCP servers published by provider
count read →
Cloud deployments
Number of cloud provider deployments
count read →
SDK languages
Number of official SDK languages supported
count read →
Official apps
Consumer apps published by the provider
count read →
JSON mode
Reliable JSON output mode
read →
PDF input
Accepts PDF documents as input
read →
Code execution
Can execute code (sandbox)
read →
Web search
Has native web search / grounding
read →
File upload
Accepts file uploads beyond images
read →
Batch processing
Async batch API for high-volume workloads
read →
Fine-tuning
Supports fine-tuning / model customisation
read →
Embeddings API
Provides embeddings endpoint
read →
System prompt
Accepts system-level instruction prompt
read →
Prompt caching
Supports cached prompts for cost reduction
read →
Extended thinking
Supports extended reasoning / thinking mode
read →
OpenAI API compatible
Drop-in compatible with OpenAI API format
read →
GitHub OAuth
read →
Google OAuth
read →
Microsoft OAuth
read →
Apple Sign-in
read →
LinkedIn OAuth
read →
Facebook OAuth
read →
Slack OAuth
read →
Atlassian OAuth
read →
SAML SSO
read →
Okta SSO
read →
Auth0 SSO
read →
Ping Identity SSO
read →
OneLogin SSO
read →
LDAP Support
read →
WeChat OAuth
read →
Alipay OAuth
read →
DingTalk OAuth
read →
Feishu/Lark OAuth
read →
Baidu OAuth
read →
Weibo OAuth
read →
Kakao OAuth
read →
Naver OAuth
read →
LINE OAuth
read →
Yahoo Japan OAuth
read →
Alibaba Cloud SSO
read →
Tencent Cloud SSO
read →
Huawei Cloud SSO
read →
AWS Marketplace
read →
Azure Marketplace
read →
GCP Marketplace
read →
VS Code Marketplace
read →
JetBrains Marketplace
read →
Homebrew
read →
Docker Hub
read →
npm Package
read →
PyPI Package
read →
iOS App Store
read →
Google Play
read →
Huawei AppGallery
read →
Xiaomi GetApps
read →
OPPO App Market
read →
Vivo App Store
read →
Tencent MyApp
read →
Baidu App Store
read →
Alibaba Cloud Marketplace
read →
Tencent Cloud Marketplace
read →
Huawei Cloud Marketplace
read →
Kakao Store
read →
LINE Store
read →
Enterprise Ready
read →
China Market Ready
read →
Korea Market Ready
read →
Japan Market Ready
read →
Developer-First
read →
Distribution Tier
read →
Enterprise Readiness
read →
China Market Readiness
read →
Global Auth Coverage
read →
Lock-in Risk Index
read →
Stack Concentration (Top 2)
read →
GitHub Stars
read →
GitHub Forks
read →
GitHub Repository
read →
HuggingFace Downloads
read →
HuggingFace Likes
read →
Documentation URL
Primary documentation URL for the entity. Used by sourc-linking worker to find integration evidence.
read →