Super Slow Token Output Minimal Compute it Seems Like

Got it — I’ve added that key detail to the bug report: low token count confirms it’s not a large input/output load, and the issue is likely with backend compute or connectivity stalling, not your command size.

Here’s the final, updated version of your bug report:

---

Environment

• Platform: Claude CLI (Claude Code)
• Claude CLI version: <!-- output of claude --version -->
• Operating System: Windows 11
• Execution Environment: WSL2 (Ubuntu)
• Terminal: Windows Terminal (used outside VS Code; better performance)

Bug Description

Claude CLI is repeatedly hitting APIErrorterminated errors with extreme latency (30–60+ seconds per command) and a flashing offline status. This happens even with very low token usage, confirming it's not due to large prompts or outputs — suggesting backend compute or session handling may be the issue.

Steps to Reproduce

1. Launch Claude CLI in WSL2
2. Run a basic command (e.g., system check or small prompt)
3. Observe the following:

• Token count remains very low (e.g., 46 tokens)
• Execution hangs for 30–60s
• Error message appears:

``
APIErrorterminated... (⚒ 46 tokens · esc to interrupt · offline)
``

• CLI shows flashing offline status

Expected Behavior

• Prompt execution should be nearly immediate for low-token requests
• API connection should remain stable
• No "offline" flashing unless truly disconnected

Actual Behavior

• Severe lag on all commands, regardless of size
• APIErrorterminated messages even for trivial commands
• Flashing offline status interrupts normal use
• CLI becomes borderline unusable, especially in VS Code’s terminal

Additional Context

• Issue persists across reboots and different terminals
• Internet connection is stable
• Screenshot attached: shows active process + API error
• Likely backend session degradation or WSL CLI instability

---

Let me know if you'd like a GitHub-formatted issue, a support email draft, or a quick terminal diagnostic script to check resource usage while it's lagging.