OpenAI SLA
99.9% uptime guarantee (Scale Tier, Enterprise (negotiated) plan only)
Official SLA Commitment
Applies to
Scale Tier, Enterprise (negotiated) plans only
Scope
Monthly uptime for Scale Tier API customers
Measurement
Service availability measured monthly; covers whether the API responds, not latency, throughput, or model output quality
Exclusions
Latency spikes, throughput degradation, model quality/accuracy, scheduled maintenance, force majeure
Not covered
Pay-as-you-go API, ChatGPT Free, ChatGPT Plus, ChatGPT Team; latency and throughput on all tiers
Promised vs Measured
SLA Promise
99.9%
≤ 43m downtime/month
Measured (30d)
99.95%
≈ 20m downtime this month
Verdict
Meeting SLA
24h Uptime
100.00%
7d Uptime
99.97%
30d Uptime
99.95%
Pricing & SLA by Tier
| Plan | SLA Guarantee | Price | Meeting 30d? |
|---|---|---|---|
| API Free Tier | No uptime SLA | $0 (usage capped at $100/mo) | — |
| API Pay-as-you-go (Tiers 1–5) | No uptime SLA | Per-token pricing, $5+ cumulative spend to unlock tiers | — |
| Scale Tier | 99.9% | From $75/unit/day (model-dependent) | |
| Enterprise Agreement | 99.9% (negotiable) | Custom ($250K+ annual spend typical) | |
| ChatGPT Plus | No uptime SLA | $20/mo | — |
| ChatGPT Team | No uptime SLA | $25/seat/mo | — |
| ChatGPT Enterprise | No uptime SLA | Custom | — |
Source: openai.com/pricing
SLA Credit Policy
Service credits applied toward future invoices as the sole remedy for uptime SLA breaches. Credits are not automatic — customers must file a claim with monitoring evidence within the designated window. Specific credit tiers are not publicly documented; Enterprise agreements typically negotiate 10–25% of monthly fees.
Industry SLA Tier Compliance
| Tier | Downtime Budget/Month | 24h | 7d | 30d |
|---|---|---|---|---|
| 99.9%(three nines)OpenAI's SLA | 43m 12s | |||
| 99.95% | 21m 36s | |||
| 99.99%(four nines) | 4m 19s | |||
| 99.999%(five nines) | 26s |
Equivalent Downtime
Last 24 hours
< 1m
Last 7 days
3m
Last 30 days
20m
Incident Summary (90 days)
Total Incidents
111
Active Now
1
Resolved
110
Avg. Resolution
7h 40m
OpenAI SLA: What You Need to Know
OpenAI guarantees 99.9% uptime measured as monthly uptime for scale tier api customers. This commitment applies exclusively to Scale Tier, Enterprise (negotiated) plan customers — other tiers have no contractual uptime guarantee.
Based on DevHelm's continuous monitoring, OpenAI currently shows a 30-day measured uptime of 99.95%. This meets their published SLA commitment of 99.9%.
What's Covered (and What Isn't)
OpenAI's SLA specifically covers monthly uptime for scale tier api customers. It explicitly excludes: Pay-as-you-go API, ChatGPT Free, ChatGPT Plus, ChatGPT Team; latency and throughput on all tiers. This means outages in excluded components don't count toward SLA credits.
Impact on Your Application
At 99.95% uptime, you can expect approximately 4h 6m of cumulative downtime per year from OpenAI alone. If your application depends on additional services, their independent failure probabilities compound — making dependency monitoring essential for maintaining your own SLA commitments.
Monitoring Methodology
DevHelm tracks OpenAI's official status page continuously and aggregates component-level status data into uptime percentages. Data is refreshed every 30 seconds. Uptime is calculated as the percentage of time all components report operational status.
