Is Claude down right now?
Authenticated API inference - 4 models monitored · How we classify outages
Weekly AI pricing & uptime digest
Price drops, new model releases, and incident summaries - every Monday. Free.
Claude is currently down - API inference failing (5+ consecutive checks) - 168ms HTTP response. Last checked . 90-day uptime: 99.9%. Claude API: 0/4 models up - check inference section below.
claude-opus-4-7 API Outage
Stay informed
Follow @tickerr_ai
We post when tier 1 LLM APIs go down - before the official status page updates.
Follow on X →Or get the weekly reliability digest:
Add a status badge
Show live Claude status on your site, README, or docs.
HTTP uptime (90d)
99.9%
20 incidents (90d)
HTTP response now
168ms
HTTP p50 (7d)
149ms
median ping response
HTTP p95 (7d)
1651ms
tail ping response
Claude API Status
Live · every 5 minBest TTFT (p50)
—
time to first token
Best throughput
—
output tokens/sec (24h avg)
Min success rate
0%
worst model (24h)
P50 = typical speed. P95 = worst case 95% of the time. Measured by Tickerr's independent inference checks. Requires ≥10 checks to display.
TTFT over 24 hours
ⓘ Authenticated streaming API calls via native fetch. TTFT = milliseconds from request start to first streamed token chunk. Throughput = output tokens ÷ generation time. Checks run from Vercel us-east-1. Independent of the provider's official status page.
Is Claude slow right now?
Claude is currently experiencing an outage or significant issues. Check the status section above for details.Last checked 2m ago.
HTTP endpoint response: 168ms (7-day p50: 149ms). This measures basic reachability, not model inference speed.
TTFT (time-to-first-token) is measured via authenticated streaming API calls every 5 minutes. Current values are compared against 7-day and 30-day rolling baselines. A minimum of 5 successful checks is required before classification. Methodology
Claude speed FAQ
Is Claude slow right now?
Claude is currently experiencing an outage or significant issues. Check the status section above for details.
Why is Claude so slow today?
Claude is currently experiencing an outage, not just slowness. Check the status section above for the latest incident details and expected resolution.
Is Claude down or just responding slowly?
Claude is currently down — not just slow. An active outage or major incident has been detected. Check the status section above for details.
How does Tickerr determine whether Claude is slow?
Tickerr sends authenticated API requests to Claude every 5 minutes and measures time-to-first-token (TTFT). Current TTFT is compared against a rolling 7-day and 30-day baseline. A model is classified as slow when its current p50 TTFT exceeds 2x its baseline, and severely slow at 3.5x. At least 5 successful checks are required before any classification is made.
What is normal response time for Claude?
Tickerr is still collecting baseline data for Claude. Normal response times will be available after sufficient monitoring history has accumulated.
Can one Claude model be slow while others are working normally?
Yes. Claude runs multiple models (e.g. claude-haiku-4-5, claude-sonnet-4-6, claude-opus-4-7), and each can have independent performance characteristics. Tickerr monitors each model separately. A slowdown on one model doesn't necessarily affect others.
Experimental capability checks
PUBLIC PILOTTickerr sends deterministic test prompts to Claude every hour and evaluates the output using code-based pass/fail checks. These results are informational onlyduring the pilot period and do not affect the tool's status verdict.
How Tickerr tests this
Every hour, Tickerr sends three deterministic test prompts to each monitored model via the provider's API. Each response is evaluated with a code-based pass/fail check — no subjective judgement.
- Instruction following: asks the model to reply with an exact word. Graded by exact string match.
- JSON schema: asks for a JSON object matching a fixed schema. Graded by parse + field validation.
- Tool call: asks the model to call a function with specific arguments. Graded by parsing the tool-call response.
Infrastructure errors (timeouts, rate limits, auth failures) are tracked separately and excluded from capability pass rates. Models that don't support tool calls are marked “not supported” rather than counted as failures. Full methodology
HTTP endpoint response time (7 days)
p50 149ms·p95 1651msⓘ HTTP response times to Claude's status endpoint - measures infrastructure availability, not API inference speed. For TTFT and model-level API status, see the Claude API Status section above.
90-day uptime comparison
Independent monitoring catches issues faster - official pages sometimes lag by hours before acknowledging problems.
Service components
via status.anthropic.com ↗Incident history
Independent monitoring detected consecutive API failures for claude-opus-4-7. Tickerr measures time-to-first-token (TTFT) via live streaming API calls every 5 minutes.
Independent monitoring detected consecutive API failures for claude-opus-4-6. Tickerr measures time-to-first-token (TTFT) via live streaming API calls every 5 minutes.
Independent monitoring detected consecutive API failures for claude-haiku-4-5. Tickerr measures time-to-first-token (TTFT) via live streaming API calls every 5 minutes.
Independent monitoring detected consecutive API failures for claude-sonnet-4-6. Tickerr measures time-to-first-token (TTFT) via live streaming API calls every 5 minutes.
We have identified an issue which resulted in some organizations to be incorrectly paused for having reached their spend limit, resulting in refused requests on the Claude API, Claude.ai, Claude Code,…
We are investigating elevated errors affecting usage data on the platform.claude.com Usage page and the Admin API usage report. Core functionality is not affected. We will provide an update as soon as…
We are investigating elevated errors loading usage data on the platform.claude.com Usage page and the Admin API usage report endpoint since 5 PM PT. Messages API requests are not affected. We will pro…
We are investigating elevated errors on requests to Claude Opus 5.5. We will provide an update as soon as possible.
We are investigating elevated errors on requests to Claude Mythos 5.1 and Claude Fable 5.1. We will provide an update as soon as possible.
We are investigating delays in credit purchases being reflected in account balances, resulting in some requests failing for insufficient credit balance. We will provide an additional update as soon as…
claude-opus-4-7 API Latency Degraded
Independent monitoring detected elevated API latency for claude-opus-4-7. Current TTFT is 4.2× above the rolling p50 baseline (3203ms vs p50 754ms). The service is responding but slower than normal. T…
We are investigating elevated error rates affecting Claude.ai (including the desktop and mobile apps), Claude Code and Claude Cowork. Users may see failed requests, errors loading or sending conversat…
claude-opus-4-7 API Latency Degraded
Independent monitoring detected elevated API latency for claude-opus-4-7. Current TTFT is 3.4× above the rolling p50 baseline (3314ms vs p50 963ms). The service is responding but slower than normal. T…
claude-opus-4-7 API Latency Degraded
Independent monitoring detected elevated API latency for claude-opus-4-7. Current TTFT is 4.1× above the rolling p50 baseline (3924ms vs p50 962ms). The service is responding but slower than normal. T…
We are investigating elevated errors on requests to Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5. We will provide an update as soon as possible.
We are aware of an issue affecting users interacting with subscriptions via Google Play, such as when creating a new subscription on Android devices. Google has confirmed an issue with Google Play, an…
We are investigating elevated errors on requests to Claude Mythos 5.1 and Claude Fable 5.1. We will provide an update as soon as possible.
claude-opus-4-7 API Latency Degraded
Independent monitoring detected elevated API latency for claude-opus-4-7. Current TTFT is 3.7× above the rolling p50 baseline (3126ms vs p50 834ms). The service is responding but slower than normal. T…
We are investigating elevated errors on requests to Claude Fable 5.1 and Claude Mythos 5.1. We will provide an update as soon as possible.
We are experiencing slower than normal response times on the Claude API for some customers whose traffic enters through the US Midwest. Users may experience elevated latency and some request timeouts.
Related pages
Claude API not working? Common error codes
If Claude's API is returning errors, the table below explains what each code means and how to fix it. If errors are widespread, check the live status above - a service incident will appear there within minutes.
| Error | What it means & what to do |
|---|---|
| HTTP 429 | Rate limit hit - check your plan tier, add exponential backoff |
| HTTP 529 (overload_error) | Anthropic is overloaded - retry after 30–60 seconds |
| HTTP 500 (api_error) | Unexpected server error - retry once, then check this page |
| HTTP 503 | Service unavailable - check for active incidents below |
Note: Tickerr monitors Claude's status endpoint, not individual API calls. An HTTP 429 or 500 in your app may be specific to your account tier - check the rate limits page for plan-specific thresholds.
About Claude status
Claude is Anthropic's AI assistant and API, available in Haiku, Sonnet, and Opus tiers. This page tracks two independent signals: HTTP availability (is Anthropic's infrastructure reachable?) and live API inference (is the Claude API actually working for developers?). Tickerr makes authenticated API calls to all six Claude models - claude-haiku-4-5, claude-haiku-3-5, claude-sonnet-4-6, claude-sonnet-4, claude-opus-4-6, and claude-opus-4 - every 5 minutes, measuring TTFT (time-to-first-token) and output throughput. This catches model-layer failures that don't show up as HTTP outages, covering roughly 35% of incidents. Common Claude API errors: HTTP 529 overloaded_error (Anthropic at capacity - retry after 30–60s), HTTP 429 rate_limit_error (you've hit plan limits - check the rate limits page), HTTP 500 api_error (unexpected server error - retry once). If the Claude API is not working or returning 529 overloaded responses, check the live status above - Tickerr typically detects these within 5 minutes. Claude API latency (TTFT) typically runs 300–800ms for Haiku models and 600ms–2s for Opus under normal load. Anthropic's official status page at status.anthropic.com is updated within 15–30 minutes of a confirmed incident; Tickerr's independent inference monitoring surfaces issues faster.
You can also check the official Claude status page.