Best AI coding plans at $20 a month or less
The entry tier, side by side, with what limits each one applies.
Data verified 2026-10-13 from official pricing pages.
Plans at $20/month or less
| Price | Plan | Vendor | Notes | Source |
|---|---|---|---|---|
| $8/mo | Cursor Plus | Cursor (sold via X Subscription Corp) | Basic pool; Grok models only | source · 2026-10-13 |
| $10/mo | GitHub Copilot Pro | GitHub / Microsoft | $15/mo of AI credits ($10 base + $5 flex) | source · 2026-10-13 |
| $20/mo | Claude Code Pro | Anthropic | $17/mo billed annually ($200/yr); includes Claude Code | source · 2026-10-13 |
| $20/mo | Claude Code Enterprise | Anthropic | $20/seat/mo + usage at API rates, billed annually | source · 2026-10-13 |
| $20/mo | Devin Pro | Cognition | source · 2026-10-13 | |
| $20/mo | OpenAI Codex Plus | OpenAI | Codex on web, CLI, IDE, iOS | source · 2026-10-13 |
| $20/mo | Windsurf (now Devin Desktop) Pro | Cognition | source · 2026-10-13 |
How each entry plan limits you
- Claude Code: Usage resets on a rolling 5-hour session window; paid plans add weekly limits. Claude Code shares one usage pool with Claude chat. Pro = at least 5x Free per session; Max = 5x or 20x Pro. No fixed message count is published. Past the limit you can enable usage credits billed at standard API rates.
- Cursor: Every tier has a usage pool shared across Cursor and Grok that refreshes weekly. Plus = Grok models only. Premium and up: third-party models (Claude, GPT, etc.) are billed separately at API list price, with optional spending caps. Top-offs available.
- OpenAI Codex: Plus: rolling 5-hour window plus possible weekly limits. OpenAI's estimate for Plus is 15-160 local GPT-6.1 Sol messages, 5-45 GPT-6 Astra, or 350-3,000 GPT-6 Luna messages per 5 hours. Pro plans currently have no 5-hour limit (weekly limits may apply). Credits can extend usage; Codex CLI can also run on an API key at API prices.
- GitHub Copilot: Agent mode, chat, code review, cloud agent and CLI consume GitHub AI Credits (1 credit = $0.01); credits vary by model. Allowance resets monthly. Completions and next-edit suggestions are unlimited on paid plans. Free: 2,000 completions + 50 chat requests. Past the allowance: set a paid-usage budget or switch to a cheaper model.
- Windsurf (now Devin Desktop): No credits: quota refreshes daily and weekly. Pro/Teams published estimates: 7-27 msgs/day on Premium Plus models (e.g. Opus 4.6, GPT-5.4), 8-101 on Premium, 47-190 on Lightweight; Max: 42-170 / 47-631 / 291-1,190. Extra usage billed at API pricing. Windsurf was rebranded Devin Desktop (third-party report); windsurf.com/pricing now 404s.
- Devin: Self-serve plans include a quota that refreshes daily and weekly; overage is billed as dollar-denominated on-demand credits at API pricing. ACUs remain for Enterprise only. Quota sizes are not published in tokens.
FAQ
What is the cheapest way to use an AI coding agent?
Among tracked plans, Cursor Plus at $8/month is the lowest-priced entry plan, but it covers Grok models only. GitHub Copilot Pro is $10/month with $15 of AI credits.
Is there a free option?
Yes: Free tiers exist for Cursor, GitHub Copilot, Windsurf, Devin and ChatGPT/Codex, each with limited agent usage. Claude Code is not included in Claude Free.
When should I pay more than $20?
When you hit weekly limits. See the next tier on the full ranking: /ai-coding-plans-compared/
Related
- Claude Code usage limits explained
- Cursor pricing 2026 (with Grok plans)
- Cheapest model for coding agents
- Codex vs Claude Code cost
- How to stop hitting AI usage limits
- Prompt caching savings calculator
- Windsurf (Devin Desktop) pricing 2026
- GitHub Copilot pricing 2026 and AI Credits
- OpenAI Codex pricing 2026
- Gemini CLI limits and pricing 2026
- Devin pricing 2026
- Every AI coding plan ranked by price
- API vs subscription: which is cheaper for agents?
- Agent cost calculator
- Cut your agent bill: 8 tactics
- All model API prices
Get price-change alerts
One short email when an AI coding agent plan, usage limit or token price changes. Free, unsubscribe anytime.
Form not loading? Subscribe on beehiiv →
