A side-by-side comparison of Claude Code and Codex — pricing, what each is best at, and whether you should pay per token instead.
Updated June 2026
Get our weekly newsletter on pricing changes, new releases, and tools.
you prefer Anthropic's models (Opus 4.8) and the most mature agentic harness.
you're in the OpenAI/ChatGPT ecosystem and want it bundled with your Plus/Pro plan.
Bottom line: Pick by ecosystem: Claude Code (Anthropic) vs Codex (OpenAI, bundled with ChatGPT). Both are top-tier CLI agents.
| Claude Code | Codex | |
|---|---|---|
| Type | CLI Agent | CLI Agent |
| Pricing model | Included in Claude Pro, Max and Ultra, or pay per token on the API | Included in ChatGPT plans (Free, Go, Plus, Pro, Business) or pay per token on the API |
| Free tier | No | Yes |
| Cheapest paid | $20/mo | $8/mo |
| Use case | Winner | Why |
|---|---|---|
| Agentic coding | Even | Both are strong terminal agents; Claude Code is more mature, Codex caught up fast on GPT-5.5. |
| Long sessions / context | Claude Code | Claude Code handles long agentic sessions and context compaction well. |
| Ecosystem / bundling | Codex | Codex is included in every ChatGPT plan, even Free and Go (desktop app) — effectively free if you already pay. |
| Price | Codex | Both start at $20 (ChatGPT Plus / Claude Pro), and Codex also has an $8 Go plan. Heavy use: ChatGPT Pro $100–500 vs Claude Max $100 / Ultra $200. |
Reddit/HN is a split decision — people pick by which model ecosystem they trust, and many keep both and route by task.
Built by @aellman
Also from the founder: BlobCap, a price index for AI training data deals, showing what AI labs actually pay for data.
2026 68 Ventures, LLC. All rights reserved.