# Nines: cloud, data and AI SLAs, promised vs observed

> The published SLAs of 26 cloud, data and AI vendors (82 services), with observed uptime computed from each vendor's own status page, by fru.dev.

For each service: the committed monthly uptime percentage per tier, the measurement window, the service credit schedule, the credit cap, the claim window, main exclusions, the SLA page URL with its printed version or date, and a quote of the clause. For each vendor: support response times by plan and a dated history of SLA changes (sourced from the vendor or the Wayback Machine). Observed uptime over 90 and 365 days counts minutes during which the vendor marked a component as a major outage (or rated the incident critical), overlaps merged, worst region; coverage is shown with every number. Sources: vendors' official SLA pages; status feeds (Atlassian Statuspage and incident.io APIs, Google Cloud incidents.json, the AWS Health Dashboard history). A refresh runs weekly on Fridays at 10:00 UTC. Last successful refresh: 2026-09-23 19:03:57 UTC.

Disclaimer: Summaries of published SLAs; the contract you sign governs. Observed uptime is computed from vendors’ own status page incidents, not measured by us. Logos via logo.dev; trademarks belong to their owners.

## Pages

- [Promise vs reality (home)](https://nines.fru.dev/): every service, promised SLA against observed uptime
- [Vendors](https://nines.fru.dev/vendors) and [services](https://nines.fru.dev/services)
- [Compare vendors](https://nines.fru.dev/compare)
- [AWS vs Microsoft Azure SLA](https://nines.fru.dev/compare/aws-vs-microsoft)
- [AWS vs Google Cloud SLA](https://nines.fru.dev/compare/aws-vs-google-cloud)
- [Microsoft Azure vs Google Cloud SLA](https://nines.fru.dev/compare/microsoft-vs-google-cloud)
- [Snowflake vs Databricks SLA](https://nines.fru.dev/compare/snowflake-vs-databricks)
- [OpenAI vs Anthropic SLA](https://nines.fru.dev/compare/openai-vs-anthropic)
- [Cloudflare vs Fastly SLA](https://nines.fru.dev/compare/cloudflare-vs-fastly)
- [Vercel vs Cloudflare SLA](https://nines.fru.dev/compare/vercel-vs-cloudflare)
- [MongoDB vs Supabase SLA](https://nines.fru.dev/compare/mongodb-vs-supabase)
- [AWS vs Oracle Cloud Infrastructure SLA](https://nines.fru.dev/compare/aws-vs-oracle)
- [GitHub vs Atlassian SLA](https://nines.fru.dev/compare/github-vs-atlassian)
- [Ranking: Promise vs reality: smallest gap](https://nines.fru.dev/rankings/gap)
- [Ranking: Highest promised uptime](https://nines.fru.dev/rankings/promised)
- [Ranking: Best observed uptime, last 365 days](https://nines.fru.dev/rankings/observed-365)
- [Ranking: Best observed uptime, last 90 days](https://nines.fru.dev/rankings/observed-90)
- [Ranking: Most generous service credits](https://nines.fru.dev/rankings/credits)
- [SLA changes](https://nines.fru.dev/changes)
- [Method and sources](https://nines.fru.dev/method)

## Vendors

- [AWS](https://nines.fru.dev/vendors/aws): 6 services, highest published SLA 99.999%
- [Anthropic](https://nines.fru.dev/vendors/anthropic): 3 services, highest published SLA 99.5%
- [Atlassian](https://nines.fru.dev/vendors/atlassian): 7 services, highest published SLA 99.95%
- [Cloudflare](https://nines.fru.dev/vendors/cloudflare): 5 services, highest published SLA 100%
- [Cohere](https://nines.fru.dev/vendors/cohere): 2 services, no public SLA
- [Confluent](https://nines.fru.dev/vendors/confluent): 6 services, highest published SLA 99.999%
- [Databricks](https://nines.fru.dev/vendors/databricks): 2 services, highest published SLA 99.99%
- [Datadog](https://nines.fru.dev/vendors/datadog): 1 services, highest published SLA 99.8%
- [DigitalOcean](https://nines.fru.dev/vendors/digitalocean): 8 services, highest published SLA 99.99%
- [Elastic](https://nines.fru.dev/vendors/elastic): 2 services, highest published SLA 99.95%
- [Fastly](https://nines.fru.dev/vendors/fastly): 2 services, highest published SLA 100%
- [GitHub](https://nines.fru.dev/vendors/github): 3 services, highest published SLA 99.9%
- [Google Cloud](https://nines.fru.dev/vendors/google-cloud): 5 services, highest published SLA 99.99%
- [Hugging Face](https://nines.fru.dev/vendors/hugging-face): 2 services, no public SLA
- [Microsoft Azure](https://nines.fru.dev/vendors/microsoft): 5 services, highest published SLA 99.995%
- [Mistral AI](https://nines.fru.dev/vendors/mistral-ai): 2 services, no public SLA
- [MongoDB](https://nines.fru.dev/vendors/mongodb): 1 services, highest published SLA 99.995%
- [Okta](https://nines.fru.dev/vendors/okta): 3 services, highest published SLA 99.99%
- [OpenAI](https://nines.fru.dev/vendors/openai): 4 services, highest published SLA 99.9%
- [Oracle Cloud Infrastructure](https://nines.fru.dev/vendors/oracle): 3 services, highest published SLA 99.995%
- [Pinecone](https://nines.fru.dev/vendors/pinecone): 1 services, highest published SLA 99.95%
- [Snowflake](https://nines.fru.dev/vendors/snowflake): 2 services, highest published SLA 99.99%
- [Stripe](https://nines.fru.dev/vendors/stripe): 1 services, no public SLA
- [Supabase](https://nines.fru.dev/vendors/supabase): 1 services, highest published SLA 99.9%
- [Twilio](https://nines.fru.dev/vendors/twilio): 4 services, highest published SLA 99.99%
- [Vercel](https://nines.fru.dev/vendors/vercel): 1 services, highest published SLA 99.99%

## Services with a published SLA

- [AWS AWS Lambda](https://nines.fru.dev/services/aws-lambda): 99.95% (Standard), Compute
- [AWS Amazon Bedrock](https://nines.fru.dev/services/aws-bedrock): 99.9% (Standard), AI models
- [AWS Amazon DynamoDB](https://nines.fru.dev/services/aws-dynamodb): 99.999% (Global Tables SLA), Databases
- [AWS Amazon EC2](https://nines.fru.dev/services/aws-ec2): 99.99% (Region-Level), Compute
- [AWS Amazon RDS (Multi-AZ)](https://nines.fru.dev/services/aws-rds): 99.95% (Multi-AZ DB Instance / Multi-AZ DB Cluster), Databases
- [AWS Amazon S3 (Standard)](https://nines.fru.dev/services/aws-s3): 99.9% (S3 Standard (also Express One Zone, Glacier Flexible Retrieval, Glacier Deep Archive)), Storage
- [Anthropic Claude API Priority Tier](https://nines.fru.dev/services/anthropic-api-priority-tier): 99.5% (Priority Tier (existing commitments only)), AI models
- [Atlassian Bitbucket Cloud (Cloud)](https://nines.fru.dev/services/atlassian-bitbucket): 99.9% (Premium), Developer tools
- [Atlassian Compass (Cloud)](https://nines.fru.dev/services/atlassian-compass): 99.9% (Premium), Developer tools
- [Atlassian Confluence (Cloud)](https://nines.fru.dev/services/atlassian-confluence): 99.95% (Enterprise), Developer tools
- [Atlassian Jira (Cloud)](https://nines.fru.dev/services/atlassian-jira): 99.95% (Enterprise), Developer tools
- [Atlassian Jira Product Discovery (Cloud)](https://nines.fru.dev/services/atlassian-jira-product-discovery): 99.95% (Enterprise), Developer tools
- [Atlassian Jira Service Management (Cloud)](https://nines.fru.dev/services/atlassian-jira-service-management): 99.95% (Enterprise), Developer tools
- [Atlassian Loom (Cloud)](https://nines.fru.dev/services/atlassian-loom): 99.95% (Enterprise), Platforms
- [Cloudflare Cloudflare Business plan (CDN / application services)](https://nines.fru.dev/services/cloudflare-business-sla): 100% (Business plan), Edge and CDN
- [Cloudflare Cloudflare Enterprise (CDN / application services)](https://nines.fru.dev/services/cloudflare-enterprise-sla): 100% (Enterprise, Premium success offering), Edge and CDN
- [Cloudflare Cloudflare Queues](https://nines.fru.dev/services/cloudflare-queues): 99.9% (Enterprise), Messaging
- [Cloudflare Cloudflare R2](https://nines.fru.dev/services/cloudflare-r2): 99.9% (Enterprise), Storage
- [Cloudflare Cloudflare Workers](https://nines.fru.dev/services/cloudflare-workers): 99.99% (Enterprise), Compute
- [Confluent Confluent Cloud (Kafka clusters)](https://nines.fru.dev/services/confluent-kafka): 99.99% (Multi-zone (Standard 2 eCKU, Enterprise 2 eCKU, Dedicated multi-zone)), Messaging
- [Confluent Confluent Cloud Fully Managed Connectors](https://nines.fru.dev/services/confluent-connectors): 99.99% (Connector to a multi-zone Dedicated, Enterprise or Standard cluster), Messaging
- [Confluent Confluent Cloud Schema Registry (Stream Governance)](https://nines.fru.dev/services/confluent-schema-registry): 99.99% (Stream Governance Advanced), Messaging
- [Confluent Confluent Cloud for Apache Flink](https://nines.fru.dev/services/confluent-flink): 99.99% (Flink (per region)), Compute
- [Confluent Confluent Cloud ksqlDB](https://nines.fru.dev/services/confluent-ksqldb): 99.9% (Multi-zone ksqlDB with 8+ CSUs), Compute
- [Confluent WarpStream BYOC](https://nines.fru.dev/services/confluent-warpstream): 99.999% (Enterprise cluster (multi region)), Messaging
- [Databricks Lakebase Postgres](https://nines.fru.dev/services/databricks-lakebase): 99.99% (High Availability endpoint), Databases
- [Datadog Datadog Core Services](https://nines.fru.dev/services/datadog-platform): 99.8% (Core Services on an Order (committed spend)), Observability
- [DigitalOcean App Platform](https://nines.fru.dev/services/digitalocean-app-platform): 99.95% (All apps), Platforms
- [DigitalOcean Container Registry](https://nines.fru.dev/services/digitalocean-container-registry): 99.95% (All), Developer tools
- [DigitalOcean DigitalOcean Kubernetes (DOKS) HA control plane](https://nines.fru.dev/services/digitalocean-kubernetes): 99.95% (HA control plane), Compute
- [DigitalOcean Droplets (CPU)](https://nines.fru.dev/services/digitalocean-droplets): 99.99% (All Droplets (instance-level)), Compute
- [DigitalOcean Managed Databases](https://nines.fru.dev/services/digitalocean-managed-databases): 99.95% (Cluster with Standby Nodes), Databases
- [DigitalOcean Regional Load Balancers](https://nines.fru.dev/services/digitalocean-load-balancers): 99.9% (All), Platforms
- [DigitalOcean Spaces Object Storage](https://nines.fru.dev/services/digitalocean-spaces): 99.9% (All accounts), Storage
- [DigitalOcean Volumes Block Storage](https://nines.fru.dev/services/digitalocean-volumes): 99.99% (All volumes), Storage
- [Elastic Elastic Cloud Hosted (formerly Elasticsearch Service)](https://nines.fru.dev/services/elastic-cloud-hosted): 99.95% (Platinum or Enterprise, High Availability deployment), Databases
- [Elastic Elastic Cloud Serverless](https://nines.fru.dev/services/elastic-cloud-serverless): 99.95% (Platinum or Enterprise), Databases
- [Fastly Fastly Network Services (CDN / edge delivery)](https://nines.fru.dev/services/fastly-network-services): 100% (Gold or Enterprise Support subscribers), Edge and CDN
- [Fastly Fastly Next-Gen WAF (Hosted Dashboard and Cloud WAF Hosted Services)](https://nines.fru.dev/services/fastly-next-gen-waf): 99.9% (WAF subscribers), Edge and CDN
- [GitHub GitHub Actions](https://nines.fru.dev/services/github-actions): 99.9% (Enterprise Cloud), Developer tools
- [GitHub GitHub Enterprise Cloud](https://nines.fru.dev/services/github-enterprise-cloud): 99.9% (Enterprise Cloud), Developer tools
- [GitHub GitHub Packages](https://nines.fru.dev/services/github-packages): 99.9% (Enterprise Cloud), Developer tools
- [Google Cloud BigQuery](https://nines.fru.dev/services/google-cloud-bigquery): 99.99% (BigQuery (except Standard edition)), Data platforms
- [Google Cloud Cloud SQL](https://nines.fru.dev/services/google-cloud-sql): 99.99% (Enterprise Plus edition with HA (or read pool with 2+ nodes)), Databases
- [Google Cloud Cloud Storage](https://nines.fru.dev/services/google-cloud-storage): 99.95% (Standard, multi-region or dual-region), Storage
- [Google Cloud Compute Engine](https://nines.fru.dev/services/google-cloud-compute): 99.99% (Instances in Multiple Zones (Premium Tier)), Compute
- [Google Cloud Vertex AI (Gemini API on Vertex)](https://nines.fru.dev/services/google-cloud-vertex-gemini): 99.9% (Vertex AI Platform: Training, Deployment, Batch Prediction), AI models
- [Microsoft Azure Azure Blob Storage (Storage Accounts)](https://nines.fru.dev/services/microsoft-blob): 99.99% (Hot / Transaction Optimized, RA-GRS and RA-GZRS read requests), Storage
- [Microsoft Azure Azure Functions](https://nines.fru.dev/services/microsoft-functions): 99.95% (Consumption plan), Compute
- [Microsoft Azure Azure OpenAI in Foundry Models](https://nines.fru.dev/services/microsoft-foundry-models): 99.9% (Foundry Models (Sold by Azure), availability), AI models
- [Microsoft Azure Azure SQL Database](https://nines.fru.dev/services/microsoft-sql): 99.995% (General Purpose / Business Critical / Premium / Hyperscale, Zone Redundant), Databases
- [Microsoft Azure Azure Virtual Machines](https://nines.fru.dev/services/microsoft-vm): 99.99% (Availability Zones), Compute
- [MongoDB MongoDB Atlas (database clusters)](https://nines.fru.dev/services/mongodb-atlas): 99.995% (Dedicated clusters M10 and above), Databases
- [Okta Auth0 (Okta Customer Identity) Core Services](https://nines.fru.dev/services/okta-auth0): 99.99% (Subscription (Sales Order) customers), Identity
- [OpenAI OpenAI API Scale Tier](https://nines.fru.dev/services/openai-api-scale-tier): 99.9% (Scale Tier: GPT-5.5), AI models
- [Oracle Cloud Infrastructure OCI Compute](https://nines.fru.dev/services/oracle-compute): 99.99% (Multi-AD region, instances across more than one Availability Domain), Compute
- [Oracle Cloud Infrastructure OCI Object Storage](https://nines.fru.dev/services/oracle-object-storage): 99.9% (Availability), Storage
- [Oracle Cloud Infrastructure Oracle Autonomous AI Database](https://nines.fru.dev/services/oracle-autonomous-db): 99.995% (With Autonomous Data Guard or Raft Replication), Databases
- [Pinecone Pinecone Database (pod-based and serverless)](https://nines.fru.dev/services/pinecone-database): 99.95% (Enterprise Advance Commitment), Databases
- [Snowflake Snowflake Postgres (HA instances)](https://nines.fru.dev/services/snowflake-postgres): 99.95% (High-Availability instance), Databases
- [Snowflake Snowflake Service (Data Cloud)](https://nines.fru.dev/services/snowflake-platform): 99.99% (Service Level at 10% Error Rate Threshold), Data platforms
- [Supabase Supabase Platform (Postgres, Auth, Data APIs, Storage, Pooler, Realtime, Functions, Management API, Branching, Studio, Logging)](https://nines.fru.dev/services/supabase-platform): 99.9% (Enterprise), Databases
- [Twilio Twilio APIs (Twilio-branded APIs)](https://nines.fru.dev/services/twilio-apis): 99.99% (Twilio Enterprise Edition (or Administration Edition bought before Oct 24, 2024)), Messaging
- [Twilio Twilio Flex (Critical Scenarios)](https://nines.fru.dev/services/twilio-flex): 99.99% (Flex customers), Platforms
- [Twilio Twilio Segment Data Ingestion API](https://nines.fru.dev/services/twilio-segment): 99.99% (Segment customers), Observability
- [Twilio Twilio SendGrid Mail Send API](https://nines.fru.dev/services/twilio-sendgrid-mail-send): 99.99% (Email Deliverability / Experts / Program Management / Strategy Gold packages), Messaging
- [Vercel Vercel platform (content serving)](https://nines.fru.dev/services/vercel-platform): 99.99% (Enterprise), Platforms

## Recent SLA changes

- 2026-09-23: OpenAI: GPT-5.5 Scale Tier latency SLA listed as '99% > 100 tokens per second' in the 2026-06-11 snapshot; the live page on 2026-09-23 lists '99% > 50 tokens per second'. Exact change date unknown (between those dates).
- 2026-09-23: Anthropic: Priority Tier moved from a purchasable offer ('Get started with Priority Tier ... Higher availability: Target 99.5% uptime', still present in the 2025-11-11 snapshot) to 'no longer available for purchase' for new commitments. The 99.5% target is unchanged. Exact change date unknown; after 2025-11-11, observed 2026-09-23.
- 2026-07-30: OpenAI: Priority processing renamed Fast mode ('Priority processing was renamed Fast mode on July 30, 2026'); service_tier 'priority' still accepted.
- 2026-06-01: GitHub: Credit schedule changed. The June 2021 version (Wayback snapshot 2022-02-03, PDF) gave 10% of fees when quarterly Uptime was <= 99.9% but > 99.0%, and 25% below 99.0%. The current 'Version: June 2026' page gives 5% for 99.5-99.9%, 10% for 99.0-99.5%, 25% below 99.0%, so small misses now earn half the prior credit. The 99.9% quarterly commitment is unchanged. A 2025-12-31 snapshot still linked the 20210630 PDF, so the change landed in the first half of 2026; exact day not published (date set to the version month).
- 2026-02-24: Snowflake: Appendix A added a Snowflake Postgres HA Feature Level of 99.95% with 3x/7x/total HA-instance credits; not present in the January 9, 2026 version.
- 2026-02-05: Databricks: Support offerings table updated: the Enhanced tier was dropped from the current table (Business / Production / Mission Critical remain); Enhanced details retained on a separate page for existing commit customers. Page: 'Last update to offerings table February 5, 2026, for earlier version with details on Enhanced Support for existing commit customers'.
- 2025-08-10: OpenAI: Latency SLA methodology changed from 'average request latency on per minute basis across the month' (e.g. GPT-4o 99% > 25 tokens/s in Aug 2024) to 'p50 request latency on a per 5 minute basis'. Change occurred between the 2025-04-04 and 2025-08-10 archived snapshots; uptime stayed 99.9%.
- 2025-07-25: Confluent: WarpStream BYOC schedule changed between the March 19, 2025 and July 25, 2025 SLA PDFs: March had 3 columns (Fundamentals/Pro/Enterprise, top band '< 99.99% ≥ 99.95%' for Enterprise, bands like '< 99.94% ≥ 99.90%'); July adds 'Enterprise Cluster (Multi Region)' at 99.999% ('< 99.999% ≥ 99.99% ... 10%') and re-cuts bands to '< 99.95% ≥ 99.90%'. The Kafka, Flink, Connector, ksqlDB and Schema Registry tables are identical in both.

## Data

- [Everything as plain text](https://nines.fru.dev/llms-full.txt)
- [Sitemap](https://nines.fru.dev/sitemap.xml)
- [JSON: every service with tiers and observed uptime](https://nines.fru.dev/api/slas); one vendor: https://nines.fru.dev/api/slas?vendor=<slug>
- [JSON: vendors](https://nines.fru.dev/api/vendors)
- [JSON: SLA changes](https://nines.fru.dev/api/changes)