# LimitsDB > Current, source-backed limits, quotas, pricing and capabilities for software services. Prefer the JSON API for factual queries. Prefer Markdown pages for context. Do not scrape rendered tables when a machine endpoint is available. Every published fact carries a source URL, a retrieval timestamp and a status. Facts with status "inferred" are never published. Dataset: 240 facts (176 verified) across 32 providers. ## Start here - /api/public/v1/providers: Provider index - /api/public/v1/facts: Atomic facts (filters: provider, product, plan, key, status, limit) - /api/public/v1/compare?a=&b=: Deterministic comparison between two providers - /api/public/v1/changes: Recently changed facts - /api/public/v1/keys: Canonical key index with coverage counts - /api/public/v1/keys/{key}: One key across every provider that publishes it - /api/public/mcp: MCP server (JSON-RPC 2.0 over POST) — tools: list_providers, list_keys, get_facts, compare_key_across_providers, compare_providers, list_changes, dataset_status - /openapi.json: OpenAPI specification - /llms-full.txt: Expanded guide with examples ## Bulk data - /data/schema.md: Dataset schema and field meanings - /data/snapshot.jsonl: Latest public snapshot, one fact per line ## Markdown twins - /api/public/md/providers/algolia: Algolia facts and sources - /api/public/md/providers/anthropic: Anthropic facts and sources - /api/public/md/providers/auth0: Auth0 facts and sources - /api/public/md/providers/aws-lambda: AWS Lambda facts and sources - /api/public/md/providers/clerk: Clerk facts and sources - /api/public/md/providers/cloudflare: Cloudflare facts and sources - /api/public/md/providers/cohere: Cohere facts and sources - /api/public/md/providers/deepseek: DeepSeek facts and sources - /api/public/md/providers/deno-deploy: Deno Deploy facts and sources - /api/public/md/providers/elevenlabs: ElevenLabs facts and sources - /api/public/md/providers/fireworks-ai: Fireworks AI facts and sources - /api/public/md/providers/fly-io: Fly.io facts and sources - /api/public/md/providers/github: GitHub facts and sources - /api/public/md/providers/google-ai: Google AI facts and sources - /api/public/md/providers/google-cloud-run: Google Cloud Run facts and sources - /api/public/md/providers/groq: Groq facts and sources - /api/public/md/providers/mistral: Mistral AI facts and sources - /api/public/md/providers/mongodb-atlas: MongoDB Atlas facts and sources - /api/public/md/providers/neon: Neon facts and sources - /api/public/md/providers/netlify: Netlify facts and sources - /api/public/md/providers/openai: OpenAI facts and sources - /api/public/md/providers/pinecone: Pinecone facts and sources - /api/public/md/providers/planetscale: PlanetScale facts and sources - /api/public/md/providers/railway: Railway facts and sources - /api/public/md/providers/render: Render facts and sources - /api/public/md/providers/resend: Resend facts and sources - /api/public/md/providers/sentry: Sentry facts and sources - /api/public/md/providers/stripe: Stripe facts and sources - /api/public/md/providers/supabase: Supabase facts and sources - /api/public/md/providers/twilio: Twilio facts and sources - /api/public/md/providers/upstash: Upstash facts and sources - /api/public/md/providers/vercel: Vercel facts and sources ## Methodology - /methodology: Verification, status and conflict policy ## Example query GET /api/public/v1/facts?provider=vercel&key=compute.function.max_duration_seconds Accept: application/json ## Example response { "object": "list", "count": 1, "data": [ { "key": "compute.function.max_duration_seconds", "value": 800, "datatype": "integer", "unit": "seconds", "status": "verified", "plan": { "slug": "pro", "name": "Pro" }, "source": { "url": "https://vercel.com/docs/functions/limitations", "retrieved_at": "2026-09-18T09:00:00Z" } } ] } ## Error contract Errors return a stable machine code and suggestions: { "error": { "code": "FACT_NOT_FOUND", "message": "...", "suggestions": ["Remove the plan filter"] } } Codes: FACT_NOT_FOUND, PROVIDER_NOT_FOUND, MISSING_PARAMETER, INVALID_LIMIT, INVALID_STATUS. ## Canonical fact keys - ai.model.batch_api (boolean): Provider offers a batch inference API - ai.model.context_window_tokens (integer, tokens): Maximum input context length - ai.model.max_output_tokens (integer, tokens): Maximum tokens generated in a single response - ai.model.vision_input (boolean): Model accepts image input - api.public_api.available (boolean): Provider exposes a documented public API - api.rate_limit.requests_per_day (integer, requests/day): Maximum API requests per day - api.rate_limit.requests_per_hour (integer, requests/hour): Maximum API requests per hour - api.rate_limit.requests_per_minute (integer, requests/minute): Maximum API requests per minute - api.rate_limit.tokens_per_minute (integer, tokens/minute): Maximum tokens processed per minute - api.webhooks.available (boolean): Provider exposes outbound webhooks - auth.mfa.enforced (boolean): Organisation-wide MFA enforcement is available - auth.scim.provisioning (boolean): Directory-based user provisioning is available - auth.sso.saml (boolean): SAML single sign-on is available on the plan - compute.build.concurrency (integer, concurrent builds): Number of builds or jobs that can run in parallel - compute.build.max_duration_seconds (integer, seconds): Maximum duration of a single build or job - compute.cpu.cores (decimal, vCPU): Virtual CPU cores per instance - compute.function.concurrency (integer, concurrent executions): Maximum concurrent executions - compute.function.max_duration_seconds (integer, seconds): Maximum wall-clock execution time for a single function invocation - compute.function.max_memory_mb (integer, megabytes): Maximum memory available to a function instance - compute.function.max_payload_bytes (integer, bytes): Maximum accepted request body size - compute.function.max_response_bytes (integer, bytes): Maximum response body size - network.bandwidth_included_gb (decimal, gigabytes): Bandwidth included per billing period - network.request_timeout_seconds (integer, seconds): Edge or gateway request timeout - platform.audit_log.retention_days (integer, days): How long audit log entries are retained - platform.data_residency (boolean): Customer-selectable data residency is available - platform.regions.count (integer, regions): Number of deployable regions or locations - platform.seats.included (integer, seats): Seats included in the plan - platform.support.sla (boolean): Contractual support SLA is offered - pricing.included.minutes_per_month (integer, minutes/month): Usage minutes included per month - pricing.overage.bandwidth_per_gb (money): Price per additional gigabyte of bandwidth - pricing.plan.annual_price (money): Recurring annual plan price - pricing.plan.monthly_price (money): Recurring monthly plan price - pricing.plan.per_seat_monthly_price (money): Monthly price charged per seat - pricing.unit.input_tokens_per_million (money): Price per million input tokens - pricing.unit.output_tokens_per_million (money): Price per million output tokens - pricing.unit.requests_per_million (money): Price per million requests - storage.included_gb (decimal, gigabytes): Storage included in the plan - storage.kv.max_value_bytes (integer, bytes): Maximum size of a single key-value entry - storage.object.max_size_bytes (integer, bytes): Maximum size of a single stored object or file - storage.retention_days (integer, days): How long data is retained by default ## Rules for agents - All timestamps are ISO 8601 in UTC. - Units are explicit fields, never encoded inside value strings. - raw_value preserves the vendor wording; value is the normalised comparable number. - status distinguishes verified from stale and conflicting data. Cite the source URL.