On this page

Status pages answer one question well: is it broken right now. They answer the next one badly: how often does it break. Ask the Claude status API for its incidents and you get the latest 50, which on 4 October 2026 reached back only to 25 July. OpenAI's returns 25, which reached back to 14 September. Ask either about last spring and the record is gone from the API.

So I started keeping it. Once a day a script reads 12 public status pages, stores each incident once under the provider and the incident's own id, and never deletes one. Where a page still exposes older history, I pulled that in too. This page is what the archive says, rebuilt every day.

Gravity's agents run on model APIs like these, which is why their incidents are the first thing I read in the morning. The numbers below are the providers' own words, counted carefully, and nothing more.

The numbers, as of 17 November 2026
  • OpenAI API: 21 incidents in the 90 days to 17 November 2026, 2 of them major or critical, adding up to 220 hours with an open incident on at least one API component.
  • Claude API: 26 incidents in the 90 days to 17 November 2026, 20 major or critical, 33 hours with an open incident, and a median of 41 min from start to resolved.
  • Most hours with an open incident among model APIs in the last 30 days: Fireworks AI, 272 hours across 85 incidents (30 days to 17 November 2026).
  • 5 model APIs posted no incident in the same 30 days: Gemini on Vertex AI (Google Cloud), Cohere, Groq, Perplexity API, Vercel AI Gateway. No posted incident is not proof of no problems; it is what the status page shows.
  • Across all 12 services tracked: 149 incidents in the 30 days to 17 November 2026, 19 of them major or critical. The archive holds 1,531 incidents from 12 status pages.

Updated . Hours with an open incident are not downtime: they measure how long the provider's own status page showed an incident as unresolved, with overlapping incidents counted once.

Cite this page: "AI API reliability report", Gravity, updated , https://gravity.fast/data/ai-api-reliability-report/. Data licensed CC BY 4.0: copy it, chart it, use it commercially, and link back here. Downloads: data.csv (one row per incident) and monthly.csv (one row per service per month).

The last 90 days

Model APIs first, then the agent and automation platforms that sit one layer up. For OpenAI, Anthropic, Perplexity and Vercel only the API part of the status page counts; the consumer apps on the same pages are left out. The coverage table further down says exactly which components count for each service.

ServiceIncidentsMajor or criticalHours with an open incidentMedian time to resolveLongest incident
Model APIs
OpenAI API2122201 h 18 min6 d 14 h
Claude API (Anthropic)26203341 min6 h 16 min
Gemini on Vertex AI (Google Cloud)000n/an/a
Cohere104.44 h 22 min4 h 22 min
Groq317.144 min6 h 5 min
Fireworks AI201033713 min10 d 3 h
Perplexity API100.637 min37 min
Vercel AI Gateway000n/an/a
Agent and automation platforms
Zapier2645052 h 51 min6 d
Make93583 h 56 min18 h 47 min
Cursor83201271 h 7 min9 h 34 min
v0 by Vercel668.01 h 3 min3 h 16 min

The 90 days to 17 November 2026. Incidents are counted by start date. Hours are the time at least one in-scope incident was open, clipped to the window. "Data from" marks a service whose usable history starts inside the window, so its figures cover less than 90 days. What counts for each service is listed under coverage. Every incident is in data.csv.

The last 30 days

The same table over a shorter window, which is the one to quote when a writer needs "recently".

ServiceIncidentsMajor or criticalHours with an open incidentMedian time to resolveLongest incident
Model APIs
OpenAI API101171 h 1 min6 h 32 min
Claude API (Anthropic)1199.740 min2 h 29 min
Gemini on Vertex AI (Google Cloud)000n/an/a
Cohere000n/an/a
Groq000n/an/a
Fireworks AI85027214 min10 d 3 h
Perplexity API000n/an/a
Vercel AI Gateway000n/an/a
Agent and automation platforms
Zapier1011493 h 4 min4 d 19 h
Make213517 h 16 min18 h 47 min
Cursor284361 h 7 min9 h 34 min
v0 by Vercel335.91 h 49 min3 h 16 min

The 30 days to 17 November 2026, same rules as the 90-day table.

Leaderboard: this month and last

Ranked by the fewest hours with an open incident, so the top of each table is the quietest status page. Early in a month these tables are thin, and a single incident can move a service several places, so the full previous month sits beneath for comparison.

Model APIs, November 2026 so far

RankServiceIncidentsMajor or criticalHours with an open incident
1Gemini on Vertex AI (Google Cloud)000
1Cohere000
1Groq000
1Perplexity API000
1Vercel AI Gateway000
6OpenAI API301.4
7Claude API (Anthropic)324.2
8Fireworks AI490226

Ranked by fewest hours with an open incident, then fewest incidents; ties share a rank. November 2026 from the 1st to 17 November 2026.

Agent and automation platforms, November 2026 so far

RankServiceIncidentsMajor or criticalHours with an open incident
1v0 by Vercel110.9
2Zapier405.6
3Cursor9113
4Make2135

Ranked by fewest hours with an open incident, then fewest incidents; ties share a rank. November 2026 from the 1st to 17 November 2026.

Model APIs, October 2026

RankServiceIncidentsMajor or criticalHours with an open incident
1Gemini on Vertex AI (Google Cloud)000
1Cohere000
1Vercel AI Gateway000
4Perplexity API100.6
5Groq317.1
6Claude API (Anthropic)11912
7Fireworks AI69070
8OpenAI API111156

Ranked by fewest hours with an open incident, then fewest incidents; ties share a rank. The full month of October 2026.

Agent and automation platforms, October 2026

RankServiceIncidentsMajor or criticalHours with an open incident
1Make101.1
2v0 by Vercel446.6
3Cursor33751
4Zapier123228

Ranked by fewest hours with an open incident, then fewest incidents; ties share a rank. The full month of October 2026.

Month by month

The last seven months for every service. This is the part no status page shows you, because their APIs forget and their history pages are one provider at a time.

ServiceMay 2026Jun 2026Jul 2026Aug 2026Sep 2026Oct 2026Nov 2026 (to date)
Model APIs
OpenAI APIn/an/a10
161 h from 6 Jul 2026
2
10.0 h
6
61 h
11
156 h
3
1.4 h
Claude API (Anthropic)29
30 h
38
455 h
39
79 h
16
28 h
8
11 h
11
12 h
3
4.2 h
Gemini on Vertex AI (Google Cloud)0
0 h
0
0 h
0
0 h
0
0 h
0
0 h
0
0 h
0
0 h
Cohere0
0 h
1
2.9 h
1
0.8 h
1
1.0 h
1
4.4 h
0
0 h
0
0 h
Groq0
0 h
0
0 h
1
1.6 h
0
0 h
0
0 h
3
7.1 h
0
0 h
Fireworks AIn/an/a24
17 h from 6 Jul 2026
135
119 h
36
21 h
69
70 h
49
226 h
Perplexity APIn/an/a0
0 h from 6 Jul 2026
0
0 h
0
0 h
1
0.6 h
0
0 h
Vercel AI Gateway0
0 h
0
0 h
1
0.5 h
0
0 h
0
0 h
0
0 h
0
0 h
Agent and automation platforms
Zapiern/a2
24 h from 18 Jun 2026
9
87 h
9
138 h
5
155 h
12
228 h
4
5.6 h
Make4
25 h
5
15 h
8
324 h
4
139 h
4
22 h
1
1.1 h
2
35 h
Cursor25
36 h
10
12 h
30
57 h
25
62 h
32
48 h
33
51 h
9
13 h
v0 by Vercel2
2.1 h
2
5.0 h
4
4.9 h
1
3.7 h
1
0.6 h
4
6.6 h
1
0.9 h

Each cell: incidents started that month, and below it hours with an open incident. Tracking since 4 October 2026, when the daily collector first ran; earlier months come from each provider's own retained history: back to 1 October 2025 for the Atlassian Statuspage pages (Claude, Make, Cursor, Vercel), about 90 days for the incident.io pages (OpenAI, Zapier, Perplexity, Cohere, Groq, Fireworks AI, v0; Cohere, Groq, Zapier and v0 reach further through their latest-25 list), and whatever Google's incidents feed still lists. "From" marks a month the history only partly covers; n/a means no usable history. Full series since 1 October 2025: monthly.csv.

How to read these numbers

Hours with an open incident are not downtime. They measure how long the provider's own status page showed at least one in-scope incident as unresolved. Most incidents are partial: slower responses, errors on one model, one region or one feature. A status page also tends to open an incident after the trouble starts and to close it when the provider is confident, so the window runs long in both directions. Treat the figure as a reporting measure.

Providers post at different thresholds. Fireworks AI posts an incident per affected model, so it logs far more incidents than a provider that groups problems into one notice. A page that posts every slowdown will look worse here than one that posts only big outages. That is a real difference in transparency, and it cuts the opposite way from how a raw ranking reads.

Scope matters more than it looks. The OpenAI and Claude rows count only incidents that touched an API component, not ChatGPT or claude.ai. The Zapier, Make, Cursor and v0 rows count every incident on the page, because each page describes one product. Zapier and Make also post incidents for single third-party connectors (one app's trigger failing, say), which can stay open for days and push their hours up even when the core product is fine. If you quote a figure, quote the scope with it.

Zero means nothing was posted. It is not a test result. If you run production traffic on any of these, the practical move is a fallback chain rather than a bet on one provider: I wrote up how to set up agent fallback chains and retry and fallback patterns, and the agent uptime and reliability guide covers what to measure on your own side. Rate limits fail differently from outages, which the rate-limiting post covers.

Change log: notable incidents

Every major or critical incident, plus anything open six hours or more, in the last 90 days, newest first. Each links to the provider's own incident page.

  1. : Claude API (Anthropic), major impact: Elevated errors for Claude Sonnet 5.5 and Claude Fable 5.2, open 1 h 15 min.
  2. : Claude API (Anthropic), major impact: Elevated errors on claude.ai, the Claude API, and Claude Code, open 2 h 29 min.
  3. : Cursor, major impact: Elevated errors for some requests, open 1 h 31 min.
  4. : v0 by Vercel, major impact: Errors sending messages in v0 for some Team and Enterprise users, open 52 min.
  5. : Make, major impact: Parts of the make.com website unavailable, open 15 h 45 min.
  6. : Fireworks AI, minor impact: Service Degradation forĀ GLM 5.3, open 20 h 5 min.
  7. : Make, minor impact: Delayed processing of incoming events, open 18 h 47 min.
  8. : Fireworks AI, no impact level posted: Service degradation, open 10 d 3 h.
  9. : Cursor, minor impact: Cloud Agent Degradation, open 7 h 42 min.
  10. : Cursor, minor impact: Investigating Service Degradation with VS Code Extensions, open 9 h 34 min.
  11. : Cursor, major impact: Multiple services degraded, open 1 h 58 min.
  12. : Cursor, major impact: Delayed responses in Grok Bot, open 10 min.
  13. : Claude API (Anthropic), major impact: Elevated errors for Claude Sonnet 5.5, open 41 min.
  14. : Claude API (Anthropic), major impact: Elevated errors for Claude Sonnet 5.5, open 42 min.
  15. : Claude API (Anthropic), major impact: Elevated errors for Claude Fable 5.1, open 32 min.
  16. : Claude API (Anthropic), major impact: Elevated errors for Claude Sonnet 5.5 and other models, open 1 h 39 min.
  17. : Claude API (Anthropic), major impact: Elevated errors for Claude Fable 5 and Claude Fable 5.1, open 37 min.
  18. : Claude API (Anthropic), major impact: Elevated errors for Claude Opus 5.5, open 6 min.
  19. : Cursor, major impact: Investigating service outage - Grok models unavailable, open 41 min.
  20. : Claude API (Anthropic), critical impact: Elevated errors for Claude Fable 5.1, Sonnet 5.5, and Haiku 5.5, open 34 min.
  21. : Zapier, major impact: Connection errors affecting Salesforce Pardot Zaps, open 3 d 12 h.
  22. : Zapier, no impact level posted: Security updates in progress for third-party software components, open 4 d 19 h.
  23. : v0 by Vercel, major impact: Delayed responses in v0, open 3 h 16 min.
  24. : OpenAI API, minor impact: Elevated ChatGPT conversation and Responses API errors, open 6 h 32 min.
  25. : v0 by Vercel, major impact: Elevated error rate on the v0 API, open 1 h 49 min.
  26. : Zapier, no impact level posted: Delayed real-time updates in Zapier Tables, open 25 h 1 min.
  27. : Zapier, no impact level posted: Upstream errors with the BambooHR integration, open 6 h 12 min.
  28. : OpenAI API, major impact: Authentication errors for some users, open 1 h.
  29. : Zapier, major impact: Errors publishing Zaps from Guided Templates with app approvals enabled, open 1 h 13 min.
  30. : Cursor, minor impact: Elevated errors affecting Grok models, open 6 h 56 min.
  31. : Tracker. Tracking started. Backfilled Claude, Make, Cursor and Vercel from their status-page history to 1 October 2025; incident.io pages (OpenAI, Zapier, Cohere, Groq, Fireworks AI, Perplexity, v0) from their 90-day windows and latest-25 lists; Google Cloud from its incidents feed.

Notable means major or critical impact, or open for 6 hours or more, in the 90 days to 17 November 2026; newest first, up to 30 lines. The page rebuilds daily, so this list moves.

Coverage and status at the last check

What counts for each service, how far back the usable history goes, how many incidents the archive holds, and what each status page said when the collector last read it.

ServiceStatus pageWhat countsUsable history fromIncidents storedStatus page at the last check
OpenAI APIstatus.openai.comIncidents touching a component in the APIs group (Chat Completions, Responses, Realtime, Batch, Embeddings and the rest)6 July 202632All clear: All Systems Operational
Claude API (Anthropic)status.claude.comIncidents touching the Claude API (api.anthropic.com) component1 October 2025299Minor issue: Partially Degraded Service
Gemini on Vertex AI (Google Cloud)status.cloud.google.comGoogle Cloud incidents that list a Vertex AI or Gemini product27 February 20261All clear: No open incident naming Vertex AI or Gemini
Coherestatus.cohere.comEvery incident on the page1 March 202525All clear: All Systems Operational
Groqgroqstatus.comEvery incident on the page3 September 202528All clear: All Systems Operational
Fireworks AIstatus.fireworks.aiEvery incident on the page (it posts per-model incidents)6 July 2026313All clear: All Systems Operational
Perplexity APIstatus.perplexity.comIncidents touching the API component6 July 20261All clear: All Systems Operational
Vercel AI Gatewaywww.vercel-status.comIncidents touching the AI Gateway component1 October 20252All clear: All Systems Operational
Zapierstatus.zapier.comEvery incident on the page18 June 202641All clear: All Systems Operational
Makestatus.make.comEvery incident on the page1 October 202560All clear: All Systems Operational
Cursorstatus.cursor.comEvery incident on the page1 October 2025256Fetch failed (HTTP 403 https://status.cursor.com/history.json?page=1); last good data kept
v0 by Vercelv0-status.comEvery incident on the page (v0 has its own status page; status.v0.dev redirects there)18 November 202530All clear: All Systems Operational

Status as read at 08:44 UTC on 17 November 2026. This is a once-a-day snapshot, not a live monitor; for "is it down right now", open the status page linked in the row.

Methodology

Here is exactly how the tables are built, so you can decide how far to trust them.

Where the data comes from

Atlassian Statuspage pages (Claude at status.claude.com, Make, Cursor, Vercel): the public /api/v2/incidents.json, /api/v2/components.json and /api/v2/summary.json endpoints every day. For the backfill and to close any gap, the page's own /history.json listing and the per-incident /incidents/<id>.json record, which carries start, resolve and component data. These pages were backfilled to 1 October 2025.

incident.io pages (OpenAI, Zapier, Cohere, Groq, Fireworks AI, Perplexity, and v0 at v0-status.com): the Statuspage-compatible /api/v2/incidents.json and /api/v2/summary.json, plus the JSON the status page itself loads, which lists every incident in its 90-day window with the affected components and impact times. The compatible API carries no component data, so components come from the second source.

Google Cloud: the public incidents feed at status.cloud.google.com/incidents.json, filtered to incidents that list a Vertex AI or Gemini product. Google's severity labels are mapped to this page's scale as service information to minor, service disruption to major, service outage to critical. That mapping is mine.

What counts

An incident is anything the provider posted as an incident. Scheduled maintenance is excluded. Each incident is stored once, keyed by status page and the provider's incident id, and later reads update it in place, for example when it resolves. Impact is the provider's own label (none, minor, major, critical) where the page gives one; for older incident.io records outside the compatible API's latest 25, it is derived from the worst component status posted (degraded performance to minor, partial outage to major, full outage to critical), a mapping that matched the provider's own label on 81 of the 83 incidents where I could compare both on 4 October 2026.

How each number is computed

Incidents are counted in the window where they started. Major or critical is the subset with that impact label. Hours with an open incident is the union of each incident's start-to-resolve interval, clipped to the window, so two overlapping incidents count once. Median time to resolve uses incidents that started in the window and have resolved. Longest incident is the longest start-to-resolve span among incidents that started in the window, linked to the provider's page. Start is the provider's stated start time where the page records one (Statuspage's started_at, the earliest component impact on incident.io), otherwise the time the incident was posted. Windows end at the time of the last collection.

Some incidents are posted after the fact, and a page that moves to a new status provider can import its old incidents with the import time as the post time. Either way the record shows a resolve time earlier than its start. Those incidents still count, dated at their resolve time, but they add no hours and no time to resolve, because the real start is unknown. The CSV flags each one.

The collector runs daily. Nothing is ever deleted from the archive, which lives in the site repository with one dated snapshot per day, so a figure quoted from this page can be checked against the day it was taken.

Known gaps

I checked 19 status pages on 4 October 2026 and kept only those with a working public JSON feed. Not included: n8n Cloud and Lindy (no Statuspage JSON at their status pages), Relevance AI and Together AI (no JSON feed found), Replit and Mistral (requests refused), OpenRouter (status JSON unavailable when checked), and the Gemini API in Google AI Studio, whose status page renders in the browser with no public JSON feed I could find. I will add any of them when a feed works.

incident.io pages expose only their last 90 days with component data, so OpenAI and Perplexity history here starts in early July 2026 and grows from there. Google's feed listed six incidents across all of Google Cloud on 4 October 2026, the oldest from 27 February 2026, so Gemini on Vertex AI has a short record. None of this measures performance from the outside: there are no synthetic pings here, only what each provider published.

Related data pages: the AI agent pricing index tracks what agent platforms charge, with dated history, and the AI agent funding tracker logs every disclosed round. For a price-first comparison, see the cheapest AI agent platforms.

FAQ

1. Is the OpenAI API down right now?

This page is not a live monitor; it reads the status pages once a day. At the last check (08:44 UTC, 17 November 2026) status.openai.com said "All Systems Operational". For a live answer open status.openai.com. Over the 30 days to 17 November 2026, 10 incidents touched an OpenAI API component, 17 hours with an open incident in total.

2. Where can I see Claude API status history?

status.claude.com shows recent incidents, and its incidents API returns only the latest 50, which on 17 November 2026 reached back to 17 August 2026. This page stores every incident that touched the Claude API (api.anthropic.com) component from 1 October 2025 onward: 299 so far, with 26 in the 90 days to 17 November 2026. The full list is in data.csv.

3. What is the most reliable LLM API?

On status-page evidence alone, over the 90 days to 17 November 2026: Vercel AI Gateway and Gemini on Vertex AI (Google Cloud) posted no incident in scope, and among those that did, the fewest hours with an open incident belonged to Perplexity API (0.6 hours across 1 incident). Treat that as a reporting record, not an uptime guarantee: providers differ in how readily they post, a narrow component list posts less, and one that posts every small slowdown looks worse than one that posts only big outages.

4. How many incidents did OpenAI and Anthropic have last month?

In October 2026, 11 incidents touched an OpenAI API component (156 hours with an open incident) and 11 touched the Claude API (12 hours). Counts come from each provider's own status page and are rebuilt daily.

5. Does Google publish Gemini API incidents?

Google Cloud's incident feed covers Gemini on Vertex AI, and this page counts incidents there that list a Vertex AI or Gemini product: 1 stored, the feed reaching back to 27 February 2026. The Gemini API in Google AI Studio has its own status page, which this page does not cover yet.

6. Are hours with an open incident the same as downtime?

No. They measure how long a provider's status page showed at least one incident as unresolved. Most incidents are partial: slower responses, errors for some models or some users, or one feature failing. A status page also opens incidents late and closes them when the provider decides, so the figure is a reporting measure, not measured downtime.

7. Why do some services show zero incidents?

Because their status pages posted none in that window. That can mean a quiet month, a narrow component list, or a provider that posts sparingly. This page compares what each provider chose to publish, so a zero is a zero on the status page, not a test result.

8. Can I reuse this data?

Yes. The tables and both CSV files (data.csv, one row per incident, and monthly.csv) are licensed CC BY 4.0: copy them, chart them, use them commercially, and credit "Gravity AI API reliability report" with a link to this page. The underlying incidents are public statements on each provider's status page.

Sources

  1. OpenAI status, incidents and components, read daily from 4 October 2026: status.openai.com
  2. Claude status (Anthropic), incidents, history and components, read daily from 4 October 2026: status.claude.com
  3. Google Cloud Service Health incidents feed and its schema, read 4 October 2026: status.cloud.google.com/incidents.json, incidents.schema.json
  4. Cohere status: status.cohere.com; Groq status: groqstatus.com; Fireworks AI status: status.fireworks.ai; Perplexity status: status.perplexity.com, all read daily from 4 October 2026
  5. Vercel status (AI Gateway component): vercel-status.com; Zapier status: status.zapier.com; Make status: status.make.com; Cursor status: status.cursor.com; v0 status: v0-status.com, all read daily from 4 October 2026
  6. Atlassian Statuspage public API reference (incidents, components, summary endpoints): developer.statuspage.io, read 4 October 2026