AI Tools · Coding
The coding market moved in October: Cursor is now sold through X Subscription, GPT-5.5 left Codex, Google stopped selling Gemini Code Assist to new customers, and Copilot added Gemini 4 Argon. Here is where each tool stands, with prices checked on October 23, 2026.
Verdict table and assistant prices last verified October 23, 2026. Specialist tool prices are linked, not quoted.
Quick answer
Claude Code (included in Claude Pro at $20) is our pick for terminal-first agent work, and OpenAI Codex (included in every ChatGPT plan, Plus at $20) is the closest alternative. GitHub Copilot (Pro $10, billed in AI Credits) is the low-friction pick inside the IDEs you already use. Cursor is now bought through X Subscription, where Premium ($30) includes Cursor Pro.
| Situation | Pick | Price (US) | Why |
|---|---|---|---|
| Terminal agent for multi-file changes | Claude Code | Included in Claude Pro ($20, or $17 annual), Max, Team, Enterprise; not on Free | Opus 5.5 on paid plans; shares the Claude 5-hour and weekly pool. |
| Agent inside the ChatGPT plan you already pay for | OpenAI Codex | Included on Free, Go, Plus ($20), Pro, Business | OpenAI estimates 15 to 160 GPT-6.1 Sol local messages per 5 hours on Plus; Pro has no five-hour limit. |
| AI inside VS Code, JetBrains and GitHub | GitHub Copilot | Free; Pro $10; Pro+ $39; Max $100; Business $19/seat | Monthly AI Credits in dollars; Gemini 4 Argon added October 13, 2026. |
| AI-native editor | Cursor via X Subscription | Plus $8 (Grok models only); Premium $30; Super $100; Ultra $200 | Premium includes Cursor Pro, SuperGrok and X Premium+ in one weekly pool. |
| Free agentic IDE | Google Antigravity | Free with basic weekly limits; higher limits with Google AI Pro and Ultra | Google points former Gemini Code Assist buyers here. |
| Hosted build, run and deploy | Replit | Core $18/mo and Pro $90/mo, billed annually | Coding, hosting and deployment in one place. |
| Agentic IDE formerly sold as Windsurf | Devin | Free; Pro $20; Max $200; Teams $80/mo + $40 per dev seat | windsurf.com/pricing now redirects to devin.ai. |
| Fast editor with light AI | Zed | Pro $10/mo | $5 of tokens included; overage at API list price plus 10%. |
Choose Claude Code if
you want an agent that reads the repo, edits files and runs commands from the terminal, and you already pay for Claude.
Choose Codex if
you already pay for ChatGPT. It is included, and Pro removes the five-hour window.
Choose GitHub Copilot if
your team lives in GitHub and mainstream IDEs and you want a dollar budget per seat.
Choose Cursor (X Premium) if
you want the editor itself rebuilt around AI and you also use Grok.
Choose Antigravity if
you want to start free, or you were a Gemini Code Assist customer looking for the replacement.
Choose Replit if
you want to go from prompt to deployed app without a local setup.
October 14: GPT-5.5 was retired from Codex with ChatGPT sign-in. Paid plans moved to GPT-6.1 Sol and Free and Go to GPT-6 Luna; API-key sessions were not affected.
October 12: Cursor’s pricing page switched to X Subscription tiers. Existing Cursor plans continue unless changed, and switching gives a prorated refund.
October 9: Google stopped selling new Gemini Code Assist Standard and Enterprise subscriptions. Existing ones auto-renew through 2026.
October 13 and October 1: GitHub added Gemini 4 Argon to Copilot and shipped Copilot computer use for desktop apps.
Both now include a coding agent in their $20 plan. Claude Code is our pick for careful multi-file edits in the terminal; Codex is the better fit if coding sits inside a broader ChatGPT workflow. On October 23, 2026 Claude Opus 5.5 was listed above GPT-6 Astra on the Artificial Analysis Intelligence Index.
GitHub Copilot Free (2,000 inline suggestions a month), Codex on ChatGPT Free (GPT-6 Luna in the desktop app) and Google Antigravity (basic weekly limits) are all free. Claude Code is not on Claude Free.
Yes. Since October 12, 2026 Cursor is sold through X Subscription: Free, Plus $8 (Grok models only), Premium $30, Super $100, Ultra $200, and Teams from $30 per user.
Cursor if you want an AI-native editor and agent-heavy loops. Copilot if you want AI inside the editor and GitHub setup you already have, with a predictable credit budget.
Yes, for well-scoped tasks with tests, human review and verification before shipping. Agents speed up the work; they do not remove the review.
Put repository rules in AGENTS.md or CLAUDE.md, use Skills and MCP only where they add clear value, and appoint one captain for the final change. The agentic coding stack explains the minimum install.
Read the agentic coding stackPricing, plan names, and product claims are checked against official vendor pages first. Recommendations are workflow-first: the winning plan is the lowest-cost option that does not bottleneck the weekly job.