Sign inJoin

Claude Code

AI Coding ToolsAI Code EditorsVibe Coding Tools
  1. About
  2. Pricing
  3. What people are saying
Visit Website

About Claude Code

Claude Code is a terminal-based coding agent designed for deep, autonomous execution on massive codebases. Powered by Sonnet 5 and the new Opus 5 model, it features a native 1-million-token context window, variable "Reasoning Effort" levels, and the ability to parallelize tasks across dozens of subagents. By default, it operates in "Auto Mode," allowing it to run commands and edit files without constant permission prompts. Developers can use "Agent View" to oversee active sessions or set up "Routines" for repetitive tasks, making it a highly capable and autonomous junior engineer.

Pricing

  • $17/mo - Pro (billed annually)
  • $20/mo - Pro (billed monthly)
  • $20/mo - Team (billed annually)
  • $25/mo - Team (billed monthly)
  • $100/mo - Max 5x tier
  • $200/mo - Max 20x tier
  • Official pricing: https://claude.com/pricing

What people are saying

With the shift to "Auto Mode" as the default permission setting, developers appreciate the speed and reduced prompt fatigue, finding the models surprisingly reliable for autonomous tasks. However, some remain cautious about letting it run completely unattended on critical professional codebases.

  • "Their newer models are heavily trained out of the 'whoopsie doopsie I deleted prod' behaviour... I've been running it for months in auto mode, heavily, getting it to do sysadmin tasks... and there's not even been a whiff of anything dumb/catastrophic." — @AussieWog93 (August 2026)
  • "I was accepting all these prompts in the manual mode for a solid hour, then I switched to Auto... Walked away from the puter, got notified it was done. It did a good job... I don't think I'd switch to Auto for a professional project... But for side projects? Yeah maybe." — @dexterlagan (August 2026)

Users still caution that its parallel subagent architecture can drain usage limits extremely quickly, with independent execution consuming large chunks of session limits.

  • "Sonnet 5 basically didn't report its thinking... worked by itself for 20 minutes and burned over 90% the 5 hour session... It made some judgement calls, no discussion or checking with me." — @Juleski70 (July 2026)
  • "Within 12 minutes, [Claude Code] ate 21% of my 5-hour limit. I am on the 5x ($100) plan... Because each parallel sub-agent reads that massive 1M context window independently, you are effectively running a dozen heavy Opus calls at the exact same time." — @TheTeddyFlame3 (July 2026)