SAN FRANCISCO — Anthropic's popular Claude AI chatbot faced scattered user reports of problems Tuesday, with complaints centering on login glitches, slow responses in Claude Code and throttled usage, even as the company's official status page declared all systems operational.

Claude
Claude

Downdetector, which tracks crowd-sourced outage reports, showed ongoing user submissions for Claude AI in the past 24 hours, though not at the level of major past incidents. Common issues included problems with Claude Chat (39% of reports), Claude Code (26%) and the mobile app (21%). Recent comments mentioned broken email token login, extended wait times for Haiku responses in the desktop app, and rapid token exhaustion on paid plans.

One user posted about being throttled since 11 a.m., while another noted that Claude Code in the desktop app took over five minutes without a response on the Haiku model. A third complained the website felt "still down in action." These echo intermittent frustrations that have surfaced throughout March 2026, a month marked by rapid feature releases and occasional reliability hiccups for the AI assistant.

Anthropic's Claude Status page at status.claude.com currently lists claude.ai, platform.claude.com, the Claude API and Claude Code as operational, with 90-day uptime figures hovering between 98.89% and 99.31%. The most recent resolved incident, posted late March 29, addressed Dispatch sessions not responding in the Claude Desktop app version 1.1.9310. Anthropic fixed an underlying issue preventing the agent from sending responses and urged users to update to version 1.1.9493.

No new "Investigating" or "Degraded Performance" alerts appeared Tuesday for core services. The API, used heavily by developers and enterprises, has maintained strong uptime near 99.04% over the past 90 days.

The complaints arrive amid heightened demand for Claude, which has seen surging popularity thanks to models like Claude Opus 4.6 — touted for advanced coding, agentic workflows and professional tasks — along with features such as computer use and multi-agent coordination. Anthropic has rolled out numerous updates in March, pushing the platform to handle more complex interactions while competing aggressively with rivals like OpenAI's GPT series.

Users on X (formerly Twitter) and Reddit have voiced similar sporadic issues in recent days. Some reported hitting usage limits unusually fast on the $100-per-month Max plan after just two active conversations, while others noted login problems specific to email token authentication (Google login reportedly still worked for some). A March 29 desktop-specific fix appeared to resolve Dispatch-related hangs for many, but isolated reports of slow or unresponsive sessions persisted into early April.

This pattern of user-reported friction without a full-blown outage mirrors earlier March incidents. On March 2, Anthropic acknowledged "elevated errors" affecting claude.ai, apps, Opus 4.6, the console and Claude Code. The company stated consumer-facing surfaces were unavailable while the API remained largely unaffected. Service was restored later that day, with Anthropic thanking users and noting efforts to scale for "incredible demand." Similar partial or elevated-error events popped up around March 11, 23–27 and 30, often tied to login paths, model-specific slowdowns or capacity strain.

Industry observers point to the challenges of scaling large language model infrastructure under explosive growth. Claude has climbed app store rankings and attracted hundreds of thousands of business customers, with valuation estimates for Anthropic exceeding $380 billion in some reports. Rapid iteration — including new agent tools, swarm coordination and features teased in leaked or discussed code — has sometimes strained backend systems, leading to throttles, partial outages or delayed responses.

Downdetector and similar trackers emphasize that spikes in reports do not always equal widespread downtime; normal daily volume can trigger alerts when frustrated users converge online. Tuesday's reports appeared geographically dispersed rather than concentrated in one region, consistent with intermittent or account-specific problems rather than a global failure.

For users encountering issues, Anthropic recommends checking status.claude.com first, then trying basic troubleshooting: refreshing the browser, clearing cache, updating the desktop or mobile app, or switching between models (Opus, Sonnet, Haiku). Paid subscribers hitting limits faster than expected may need to monitor usage or contact support. The company has historically responded to major incidents with targeted fixes and occasional credits for affected API users.

Anthropic has not issued a new public statement specifically addressing Tuesday's user reports as of late evening KST. The firm continues to emphasize reliability improvements alongside feature velocity, including enhancements for coding agents and enterprise deployments.

Claude's March 2026 has been a tale of innovation and growing pains. The company shipped updates for Opus 4.6, expanded agent capabilities and integrated tools that let users orchestrate multiple AI "teammates." At the same time, uptime for claude.ai dipped to 98.25% for the month in some tracking, below the near-perfect figures of prior periods. Analysts say such fluctuations are common as frontier AI labs balance cutting-edge releases with the massive compute demands of millions of concurrent users.

Competitors have faced similar complaints. OpenAI, Google and others have weathered their own outage waves and rate-limit frustrations during periods of hyper-growth. For Claude users, the mix of powerful capabilities and occasional hiccups has become a familiar trade-off.

Those still unable to access Claude can try the API directly if they have developer access, as it has shown higher resilience in past incidents. Free-tier users sometimes experience stricter limits or earlier impacts during capacity squeezes.

As AI assistants become daily tools for coding, writing, research and automation, even brief disruptions draw quick attention on social media. Hashtags like #ClaudeDown occasionally trend during spikes, amplifying individual experiences.

Anthropic's team has a track record of transparent status updates and quick resolutions once root causes are identified. Tuesday's situation appeared more fragmented — a collection of login quirks, response delays and usage complaints — rather than the synchronized global outages seen earlier in the month.

Looking ahead, Anthropic is expected to keep investing heavily in infrastructure to support its growing user base and ambitious roadmap. Features like enhanced multi-agent systems and deeper tool integration promise more value but also require robust scaling.

For now, the majority of Claude services remain accessible, according to official monitoring. Users experiencing persistent problems are encouraged to report details via Downdetector or Anthropic's support channels to help the team pinpoint any lingering edge cases.

Claude AI, built by Anthropic, offers conversational intelligence through models including Opus, Sonnet and Haiku. It serves both casual users on claude.ai and enterprises via API, powering tasks from simple queries to complex coding and agent-driven workflows.

As of Tuesday evening, no major new incident had been posted on the status page. The platform's 90-day performance remains solid overall, though March's rapid pace has tested its limits.

Users are advised to stay updated via status.claude.com and Anthropic's official channels for the latest