Sudden usage limit spike from 38% to 100% in 20-30 seconds on fresh sessions

Bug Description
Over my last two sessions, my usage has inexplicably jumped from 38% to 100% within 20–30 seconds. This issue appeared overnight and is completely halting my workflow. I am a regular, consistent user who prefers Claude for its superior experience, but this kind of sudden, severe limitation makes relying on the service highly doubtful. I am certain this is a backend issue on your end.

Steps to Reproduce

Start a completely new session.

Apply standard prompt and context optimizations.

Monitor the API/service usage via an active usage counter.

Observe the usage jumping abruptly from standard levels (e.g., 38%) to 100% within 20–30 seconds of activity.

Expected Behavior

Usage limits should deplete linearly and accurately based on the actual tokens processed in the current session.
Actual Behavior

Usage limit spikes instantly to 100%, locking out further prompts. I witnessed this happen in real-time on my usage counter twice in a row today.

Environment & Context

Session State: Fresh session. The context window was under 12%, so there was no need for conversation compaction.

Background Processes: None. There are no parallel operations, unnecessary tools, or background connectors running.

Optimization: All standard optimizations are active and functioning correctly.

Impact

My work is being severely hampered. I cannot maintain a standard workflow if limits are falsely triggered instantly on a fresh, low-context session. Please investigate the backend routing or token-counting mechanisms to resolve this anomaly.

Environment Info

• Platform: linux
• Terminal: vscode
• Version: 2.1.143
• Feedback ID: 28177a74-dc92-4f88-92a7-f42b722b7da6
• Model: Opus 4.7 | Ctx Used: 8.0%

Errors

[{"error":"Error: 429 {\"type\":\"error\",\"error\":{\"type\":\"rate_limit_error\",\"message\":\"This request would exceed your account's rate limit. Please try again later.\"},\"request_id\":\"req_011CbA5pSwGFBeH8mCJUYnRc\"}\n    at generate (/$bunfs/root/src/entrypoints/cli.js:11:62268)\n    at makeRequest (/$bunfs/root/src/entrypoints/cli.js:51:7694)\n    at processTicksAndRejections (native:7:39)","timestamp":"2026-05-18T13:06:07.268Z"},{"error":"Error: 429 {\"type\":\"error\",\"error\":{\"type\":\"rate_limit_error\",\"message\":\"This request would exceed your account's rate limit. Please try again later.\"},\"request_id\":\"req_011CbA7CXfFr7QJZScf6Vxff\"}\n    at generate (/$bunfs/root/src/entrypoints/cli.js:11:62268)\n    at makeRequest (/$bunfs/root/src/entrypoints/cli.js:51:7694)\n    at processTicksAndRejections (native:7:39)","timestamp":"2026-05-18T13:24:13.377Z"}]