Data directories
AI developer data directories
43 dated, primary-sourced directories for developers building with AI models and coding agents: prices, limits, capabilities, shutdown dates, local models, CLI flags, MCP support and API errors. Every record links to the first-party page it came from, and each directory shows the date it was last verified.
Models and APIs
24 directoriesHosted model prices, specs, limits, benchmarks and shutdown dates, plus how each API behaves and the retrieval and gateway layers around them.
- ModelsAPI prices and specs167 models
- Model API capabilitiesTool calling, structured outputs, vision, caching, batch and reasoning per model, with sources7 capabilities
- AI model context windowsContext window and max output for every model, largest first, with long-context price thresholds and multipliers
- AI model knowledge cutoff datesProvider-stated knowledge cutoffs with release dates, the gap between them and sources66 stated cutoffs
- Free LLM API tiersWhich LLM APIs have a standing free tier, credits or none, with free limits, card rules and whether each trains on your data, quoted from the provider's terms17 providers
- LLM API cost leversPrompt caching, batch discounts, service tiers and other discounts for 15 LLM API providers, with verbatim quotes and per-model cached and batch prices15 providers
- AI API data retention and residencyHow long 18 AI API providers keep your prompts, whether they train on them, who offers zero data retention and which regions they process in, with verbatim quotes18 providers
- Best models by use caseModels ranked by one named, dated benchmark per use case6 use cases
- Embedding modelsDimensions, max input and price per 1M tokens59 models
- Reranker modelsRerank API prices for 42 models from 10 providers in each provider's own billing unit (per 1M tokens, per 1,000 queries, hourly), never converted, with context, document limits, open weights and verbatim quotes42 models
- Vector databases24 vector databases (Pinecone, Qdrant, Weaviate, Milvus, pgvector, S3 Vectors, turbopuffer and more): prices in each product's own unit, never converted, max dimensions, hybrid search, built-in embedding and reranking, licenses and recent renames, with verbatim quotes24 products
- Speech AI APIsSpeech-to-text, text-to-speech and realtime voice pricing for 79 models from 12 providers, in each provider's own unit, with verbatim quotes and retirement dates79 models
- Image generation API pricingImage generation API prices for 74 models from 12 providers, per image by size and quality, per 1M image tokens, per megapixel or credits, never converted, with verbatim quotes and retirement dates74 models
- Video generation API pricingVideo generation API prices for 64 models from 14 providers, per second by resolution, per video, per 1M tokens, credits or Units, never converted, with verbatim quotes, retirement dates and where the docs disagree64 models
- Fine-tuning APIsWhich of 12 providers still offer fine-tuning, by method, with training prices grouped by unit, hosting requirements, weight export and OpenAI's wind-down dates, quoted from the docs12 providers
- Open-weight model hosting pricesWhere to run 12 open-weight models (DeepSeek V4, Kimi K3, GLM-5.3, Qwen3.8, MiniMax-M3, gpt-oss and more) through 18 inference providers: per-token prices, cached and batch rates, context served, quantization and variants, with verbatim quotes12 models
- Model release timelineEvery dated API model release
- BenchmarksSourced leaderboards16 benchmarks
- Rate limitsLimits by provider and tier19 providers and plans
- DeprecationsModel shutdown dates11 providers
- OpenAI-compatible and Anthropic-compatible API endpointsBase URLs, endpoints, auth and documented limitations of the OpenAI-compatible and Anthropic-compatible layers per AI API provider, from official docs20 providers
- LLM gateways and routers18 LLM gateways and routers (OpenRouter, Vercel AI Gateway, Cloudflare, Portkey, LiteLLM, Kong and more): fees in each vendor's own words, BYOK, OpenAI and Anthropic endpoints, fallbacks, caching, spend controls and licenses, with verbatim quotes18 products
- LLM observability and evaluation tools20 LLM tracing and eval tools (Langfuse, LangSmith, Braintrust, Arize Phoenix, Datadog, Promptfoo and more): who changed hands, prices in each vendor's own unit, OpenTelemetry ingest, evals, self-hosting and compliance, with verbatim quotes20 products
- AI API error referenceDocumented error codes for 12 AI API providers with causes, fixes and retry guidance204 errors
Local and open weights
4 directoriesOpen-weight models, the runtimes that serve them, the hardware they fit on and the licenses they ship under.
- Local modelsRun open weights yourself63 models
- Local LLM runtimesOllama, LM Studio, llama.cpp, vLLM and more compared feature by feature12 runtimes
- Local LLM hardwareWhich open-weight models fit in the memory of 33 NVIDIA, AMD and Apple GPUs and Macs, with vendor specs33 devices
- Open-weight model licensesWhich open-weight LLMs allow commercial use: thresholds, attribution, naming rules, output restrictions and gating, with verbatim license quotes11 models
Coding agents and tools
12 directoriesSetup steps, error fixes, CLI flags, instruction files, official SDKs, agent frameworks, code sandboxes and MCP.
- Setup guidesModels in your coding tools125 guides
- Error fixesCoding agent errors, solved132 errors
- CLI referenceAgent flags and commands8 CLIs
- Coding agent instruction files matrixWhich coding agents read CLAUDE.md, AGENTS.md, GEMINI.md and rules directories, with user-level files, nesting and imports14 agents
- Official AI API SDKs by languageWhich AI API providers ship an official SDK for Python, TypeScript, Go, Java, C#, Ruby, PHP, Rust, Swift and Kotlin, with package names and install commands19 providers
- AI coding tool plans and usage limitsPlans, prices and usage limits for 18 AI coding tools (Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Devin, Kiro and more): what changed, prices in each vendor's own terms, limit windows, overage and BYOK, with verbatim quotes18 tools
- AI agent frameworksLangGraph, CrewAI, OpenAI Agents SDK, Google ADK and more: languages, versions, MCP, durable execution and other documented features16 frameworks
- AI agent code sandboxesE2B, Daytona, Modal, Vercel Sandbox, cloud primitives and model-provider code execution tools: isolation, session limits, auto-pause, egress controls and pricing in each provider's own unit, never converted, with verbatim quotes20 sandboxes
- AI agent memory services and frameworks15 agent memory products (Mem0, Zep, Letta, Supermemory, Cognee, Google Memory Bank, AgentCore Memory, the Claude memory tool and more): what changed, prices in each product's own unit, knowledge graphs, MCP servers, self-hosting and deletion, with verbatim quotes15 products
- Voice AI agent platforms16 voice agent platforms and APIs (Vapi, Retell, Bland, ElevenLabs Agents, OpenAI Realtime, Gemini Live, LiveKit Agents, Pipecat and more): Hume EVI's shutdown and other changes, prices in each vendor's own unit, telephony, model choice and compliance, with verbatim quotes16 products
- MCP serversOfficial servers, verified
- MCP client support matrixWhich apps support MCP tools, resources, prompts, sampling, elicitation, transports and OAuth24 clients
Comparisons and terms
3 directoriesHead-to-head comparisons of tools and models, and the vocabulary behind them.
Newsletter
Know when the data changes
One email per week. New model releases, pricing, limit and shutdown changes, and what they mean for what you build.