Claude Code vs Codex pricing starts at the same number. Claude Pro is $20/mo and includes Claude Code. ChatGPT Plus is $20/mo and includes Codex. The limits are what differ. Claude Code gives you a 5-hour session allowance and a weekly allowance, shared with claude.ai, and Anthropic does not publish either as a message count. Codex publishes a local message range per 5-hour window for each model, and says weekly limits may also apply.
September 2026 changed both sides. Anthropic's 50% weekly-limit promotion ended on September 13. On September 14 the weekly cap settled 25% above the old baseline, which Anthropic itself describes as a 17% cut versus the summer. OpenAI paused new ChatGPT Pro $200 sign-ups on September 10 because of GPT-6 Astra demand, then added GPT-6 Sol and Luna to Codex on September 22. Anthropic answered the same day. Claude Opus 5.5 shipped, 5-hour limits went up on Pro, Max, Team, and seat-based Enterprise, and Claude Code now defaults to Opus on every subscription plan.
This page is the plan decision guide: 5-hour windows, weekly caps, credit rates, and which subscription fits your usage. For the capability side (benchmarks, subagent architecture, failure modes), see the Codex vs Claude Code comparison.
Which Plan Should You Buy?
Both tools start at $20/mo. The right plan depends on how much you use it, which model you run, and whether your work is long interactive sessions or fanned-out cloud tasks.
| Your usage | Best plan | Cost | Why |
|---|---|---|---|
| Light, a few hours/week | Claude Pro or ChatGPT Plus | $20/mo | Both entry tiers cover light daily coding |
| Long continuous coding sessions | Claude Code on Pro/Max | $20-$200/mo | Budget depends on work done, not message count |
| Many parallel cloud tasks + GitHub reviews | Codex on Plus/Pro | $20-$100/mo | Pick a cheaper model (Sol, Luna) to stretch the window |
| Heavy daily use, one tool | Claude Max 5x or ChatGPT Pro $100 | $100/mo | 5x the entry tier on both |
| Maximum throughput | Claude Max 20x | $200/mo | ChatGPT Pro $200 (20x) closed to new sign-ups since Sept 10 |
| Pay only for what you run | API key on either | Per-token | No cloud features in Codex API-key mode |
What Changed in September 2026: Claude Code vs Codex Limits
Most Claude Code vs Codex limits pages were written before these dates. Each one changes the math.
| Date | Tool | Change |
|---|---|---|
| May 13 - Sep 13, 2026 | Claude Code | Temporary 50% boost to weekly limits on Pro, Max, Team, seat-based Enterprise. 5-hour limits untouched. |
| Aug 31, 2026 | Codex | GPT-5.4 and GPT-5.4 mini removed from Codex for ChatGPT sign-in. API keys still work. |
| Sep 3, 2026 | Codex | GPT-6 Astra launches in ChatGPT and Codex |
| Sep 10, 2026 | Codex | New sign-ups and upgrades to ChatGPT Pro $200 (20x) paused. Pro $100, Plus, Go, API unaffected. |
| Sep 14, 2026 | Claude Code | Weekly limits permanently 25% above pre-promotion baseline. Net 17% below the promotion. |
| Sep 14, 2026 | Codex | GPT-5.5 retirement announced: it leaves ChatGPT and Codex on Oct 14, 2026. GPT-5.3-Codex-Spark removed. |
| Sep 22, 2026 | Codex | GPT-6 Sol and Luna added to Codex for Plus, Pro, Business, Enterprise, Edu. Codex CLI 0.156.0 turns /usage into an analytics dashboard and turns voice on by default. |
| Sep 22, 2026 | Claude Code | Claude Opus 5.5 launches ($4/$20 per MTok on the API). Anthropic raises 5-hour limits on Pro, Max, Team, seat-based Enterprise and grants a saveable limit reset. |
| Sep 22, 2026 | Claude Code | v2.1.280: Opus 5.5 is the default Opus. Pro and Team Standard switch from Sonnet to Opus by default, so every subscription plan now starts on Opus 5.5. |
The Claude math: if the original weekly cap was 100 units, summer users had 150 and now have 125. The Codex math: GPT-6 Astra allows 5-45 messages per window on Plus, under a third of GPT-6 Sol's 15-150. Switching the Codex model changes your effective plan size more than switching plans does.
How Each Tool Meters Usage
The unit of metering is the most important difference for picking a plan. They do not compare one to one.
| Dimension | Claude Code | OpenAI Codex |
|---|---|---|
| Short-term window | Session limit, resets every 5 hours | 5-hour window, local message range per model |
| What counts | Model, conversation length, tool use, effort level | Messages, converted to credits by model and token count |
| Long-term cap | Weekly limit across all models | "Weekly limits may also apply"; none on Enterprise/Edu flexible pricing |
| Shared across surfaces | Claude and Claude Code share one limit | Local messages and cloud chats share one allowance |
| Over the limit | Wait, spend a saved limit reset, buy usage credits, move up a tier, or use an API key | Buy extra credits on Plus/Pro, redeem a banked reset, move up a tier, or use an API key |
| Check usage in the CLI | /cost shows session spend on API billing; plan limits and resets live in Settings > Usage | /status, or /usage (analytics dashboard since CLI 0.156.0) |
The 5-Hour Window, Explained
Both tools reset a short-term allowance every 5 hours. What fills it is where they diverge.
Claude Code: a session budget, not a message count
Anthropic describes the Max plan as a session-based usage limit that resets every five hours, with Max 5x at five times Pro's per-session allowance and Max 20x at 20 times. It does not convert that to prompts. Anthropic says usage varies with conversation length, model choice, tool use, and effort level. An Opus 5.5 session with high effort and many tool calls drains the window faster than a short Sonnet session. That matters more now: since v2.1.280 (September 22, 2026) Claude Code starts Pro and Team Standard users on Opus instead of Sonnet. The 5-hour limits were not part of the weekly promotion that ended September 13, so they did not drop. On September 22 Anthropic raised them on Pro, Max, Team, and seat-based Enterprise with the Opus 5.5 launch. It did not publish the size of the increase.
Codex: messages per model, cloud included
Codex publishes a range per model. On ChatGPT Plus: GPT-6 Astra 5-45 local messages per 5 hours, GPT-6 Sol 15-150, GPT-5.6 Terra 25-200, GPT-5.5 15-80, GPT-6 Luna 350-3,000. Cloud chats share the same allowance, so a heavy cloud run eats into your local budget. Code review only counts when Codex runs GitHub reviews. GPT-5.4 and GPT-5.4 mini left Codex for ChatGPT sign-in on August 31, 2026, and GPT-5.5 follows on October 14.
Limit resets on both sides
Both vendors now hand out banked resets. With Opus 5.5, Anthropic gave subscribers a limit reset they can save. One reset refills the 5-hour or the weekly limit. You redeem it with "Reset for free" in Settings > Usage on the web or Claude Desktop, not from the Claude Code terminal, and it can expire. Claude Code v2.1.234 also added "Continue automatically at usage limit", which resumes work when the limit resets. Codex has banked rate-limit resets for Plus and Pro since June 2026, and its CLI rate-limit banner links to them.
Codex GPT-6 Astra Messages per 5-Hour Window
Upper end of OpenAI's published range. Higher is more headroom before the cap.
GPT-6 Sol ranges on the same plans: 15-150, 70-700, 300-3,000. Source: OpenAI Codex pricing, September 2026.
Claude Code Weekly Caps by Plan
Claude Code adds a weekly limit on top of the 5-hour session limit. Anthropic publishes multipliers, not hours. Treat any page that quotes a fixed number of Sonnet or Opus hours per week as a third-party estimate.
| Plan | Price | Usage vs Pro | Weekly cap |
|---|---|---|---|
| Pro | $20/mo ($17/mo annual) | 1x | Yes, all models |
| Max 5x | $100/mo | 5x per session | Yes, all models |
| Max 20x | $200/mo | 20x per session | Yes, all models |
| Team premium seat | $100/seat/mo annual ($125 monthly) | 5x a standard seat | Yes |
Since September 14, 2026, weekly caps on Pro, Max, Team, and seat-based Enterprise are 25% above their pre-May baseline.
Codex Messages, Cloud Tasks, and Credit Rates
Codex meters messages per 5-hour window, then converts model usage into credits. Plus and Pro users who hit the limit can buy additional credits instead of upgrading.
| Model | Plus $20 | Pro 5x $100 | Pro 20x $200 |
|---|---|---|---|
| GPT-6 Astra | 5-45 | 25-225 | 100-900 |
| GPT-6 Sol | 15-150 | 70-700 | 300-3,000 |
| GPT-6 Luna | 350-3,000 | 1,750-14,000 | 7,000-56,000 |
| GPT-5.6 Sol | 10-100 | 50-500 | 200-2,000 |
| GPT-5.6 Terra | 25-200 | 125-1,000 | 500-4,000 |
| GPT-5.6 Luna | 250-2,000 | 1,250-10,000 | 5,000-40,000 |
| GPT-5.5 (retires Oct 14) | 15-80 | 75-400 | 300-1,600 |
Standard Business seats match Plus ranges. GPT-5.4 and GPT-5.4 mini still appear on OpenAI's table but left Codex for ChatGPT sign-in on August 31, 2026. Enterprise and Edu on flexible pricing have no fixed rate limits. API-key users pay per token and get no cloud features.
Codex credit rates per 1M tokens
| Model | Input | Cached | Output |
|---|---|---|---|
| GPT-6 Astra | 250 | 25 | 1,250 |
| GPT-6 Sol | 50 | 5 | 250 |
| GPT-6 Luna | 2.5 | 0.25 | 12.5 |
| GPT-5.6 Sol | 100 | 10 | 500 |
| GPT-5.6 Terra | 50 | 5 | 300 |
| GPT-5.6 Luna | 5 | 0.5 | 30 |
| GPT-5.5 | 125 | 12.5 | 750 |
GPT-5.6 usage averages 5-30 credits per message. Fast mode costs 2.5x the standard credit rate on GPT-6 Astra, Sol, and Luna. GPT-6 Astra output costs 5x GPT-6 Sol per token, which is why its message range is under a third of Sol's. Credit rates are 25x the API dollar price, with no cache-write charge. Source: OpenAI Codex pricing.
Plan-by-Plan Pricing
| Tier | Claude Code (via Claude) | Codex (via ChatGPT) |
|---|---|---|
| Free | Not included on free Claude.ai | ChatGPT Free $0 (quick tasks) |
| Entry paid | Pro $20/mo ($17/mo annual) | Plus $20/mo (Go $8/mo for lightweight tasks) |
| Heavy | Max 5x $100/mo | Pro 5x $100/mo |
| Maximum | Max 20x $200/mo | Pro 20x $200/mo (new sign-ups paused) |
| Team | Team $20/seat annual ($25 monthly); premium $100/$125 | Business $20/user annual ($25 monthly) |
| Pay-per-token | Console (API) account; Opus 5.5 $4/$20 per MTok | OpenAI API key (no cloud features) |
The ChatGPT Pro $100 tier launched April 9, 2026. Since September 10 it is the only Pro tier open to new buyers. OpenAI has not given a date for reopening $200 sign-ups. See Morph pricing if you want to run open-source models per token instead of paying for a subscription window.
Install and Sign In
Claude Code
Native install (recommended). macOS, Linux, and WSL:
curl -fsSL https://claude.ai/install.sh | bashWindows PowerShell:
irm https://claude.ai/install.ps1 | iexAlso available via brew install --cask claude-code (stable channel) or claude-code@latest, winget install Anthropic.ClaudeCode, signed apt/dnf/apk repos, and npm install -g @anthropic-ai/claude-code. The npm package needs Node.js 22 or later as of v2.1.198. Native installs auto-update; set "autoUpdatesChannel": "stable" to stay about a week behind latest. Requirements: macOS 13.0+, Windows 10 1809+, Ubuntu 20.04+, Debian 10+, Alpine 3.19+, 4 GB+ RAM. The current release is v2.1.280 (September 22, 2026), which defaults to Claude Opus 5.5 on every subscription plan. Opus 5.5 needs 2.1.280, which reached the latest channel first. Stable-channel and Homebrew claude-code users may see it greyed out until their channel catches up.
OpenAI Codex
Mac and Linux:
curl -fsSL https://chatgpt.com/codex/install.sh | shWindows:
powershell -ExecutionPolicy ByPass -c "irm https://chatgpt.com/codex/install.ps1 | iex"Also npm install -g @openai/codex or brew install --cask codex. Run codex and choose "Sign in with ChatGPT." Surfaces: the CLI, IDE extensions for VS Code, Cursor, and Windsurf, a desktop app (codex app; since July 9, 2026 Codex lives inside the ChatGPT desktop app), and Codex Web at chatgpt.com/codex. Codex CLI 0.156.0 (September 22, 2026) turns /usage into an analytics dashboard with account usage, token totals, and plugin and skill activity. It also adds an optional fullscreen /tui, turns voice conversations on by default (F8 toggles), and enables worktree support by default. Run codex --model gpt-6-sol to start on the cheaper GPT-6 model.
MCP support
Both support Model Context Protocol servers. Claude Code v2.1.280 added CLAUDE_CODE_MAX_MCP_DESCRIPTION_LENGTH, which caps each MCP tool description (default 2,048 characters). Lowering it trims the tool schema sent on every turn, which saves context and, on a subscription, some of your window.
If You Run Models Yourself Instead
Subscription caps exist because frontier models are expensive to serve. If your workflow is API-driven, or you want open-source models with no 5-hour window or weekly cap, the numbers that matter are per-token cost and output fidelity.
Morph serves DeepSeek with 16-bit (bf16) activations and no fp8/int8 activation quantization. Many serverless providers quantize activations to fp8 to cut cost, which shifts outputs away from the reference weights. Morph also runs codegen-specific speculative decoding and custom inference kernels tuned for coding agents.
| Option | Input / 1M | Output / 1M | Notes |
|---|---|---|---|
| morph-dsv4flash (DeepSeek V4 Flash) | $0.141953125 | $0.399625 | 16-bit activations, codegen-tuned |
| Codex GPT-6 Astra (credit-based) | 250 credits | 1,250 credits | Draws from the 5-hour window |
| Claude Opus 5.5 (API) | $4.00 | $20.00 | Default Opus in Claude Code v2.1.280 |
See Morph models and pricing for the full list.
Decision Framework
| Your situation | Buy this | Reason |
|---|---|---|
| Solo dev, a few hours a day | Claude Pro $20 or ChatGPT Plus $20 | Entry tier covers light daily coding either way |
| Long continuous pairing sessions | Claude Code (Pro, then Max) | Session budget scales 5x and 20x with Max |
| Background cloud tasks + GitHub review | Codex (Plus, then Pro $100) | Cloud and local share one allowance; pick Sol or Luna to stretch it |
| Need the cheapest paid Codex tier | ChatGPT Go $8/mo | Lowest Codex entry price, lightweight tasks |
| Want a $200 tier today | Claude Max 20x | ChatGPT Pro $200 is closed to new sign-ups since Sept 10 |
| Hit the cap mid-task often | Move up a tier or buy credits | Max 5x / Pro $100, or extra Codex credits on Plus/Pro |
| No usage window at all | API key, or run open models on Morph | Per-token billing, no 5-hour or weekly cap |
Frequently Asked Questions
What is the 5-hour limit in Claude Code and Codex?
Both tools reset a usage allowance every 5 hours. Claude Code's session limit is shared with claude.ai and is not published as a message count; it depends on model, conversation length, tools, and effort level, and a weekly limit across all models sits on top. Codex publishes local message ranges per 5-hour window for each model. On ChatGPT Plus that is 15-150 GPT-6 Sol messages or 5-45 GPT-6 Astra messages, and cloud chats draw from the same allowance. Anthropic raised Claude's 5-hour limits on September 22, 2026, without publishing a number.
Which plan should I buy for Claude Code vs Codex?
For light use, both start at $20/mo: Claude Pro includes Claude Code, ChatGPT Plus includes Codex. For heavy use, Claude Max 5x is $100/mo with 5x Pro's per-session usage, and ChatGPT Pro $100 gives 5x Plus's Codex usage. At the top, Claude Max 20x is $200/mo. ChatGPT Pro $200 (20x) is closed to new sign-ups since September 10, 2026, so new buyers can only get Codex's 5x tier.
Is Codex cheaper than Claude Code?
At list price they match: $20, $100, and $200 tiers on both sides, and Codex adds Go at $8/mo. What differs is work per dollar. Codex on Plus allows 15-150 GPT-6 Sol messages per 5 hours but only 5-45 on GPT-6 Astra, so model choice changes your effective price by about 3x. One Hacker News user on September 22, 2026 estimated about $200 of API-equivalent usage per week on Codex Pro $100 versus about $5,000 on Claude Max $200.
Does Codex have usage limits like Claude Code?
Yes. Codex has a 5-hour window with per-model local message ranges, and OpenAI's pricing page says weekly limits may also apply. Only Enterprise and Edu on flexible pricing have no fixed rate limits. Plus and Pro users who hit the limit can buy extra credits instead of upgrading. Claude Code has a 5-hour session limit plus a weekly limit across all models on Pro, Max, Team, and seat-based Enterprise. Past the limit, Claude users can buy usage credits (formerly extra usage) or spend a saved limit reset.
Did Claude Code weekly limits change in September 2026?
Yes. Anthropic ran a 50% weekly-limit boost from May 13 through September 13, 2026. From September 14, weekly limits are permanently 25% higher than before the promotion on Pro, Max, Team, and seat-based Enterprise. Anthropic's own framing: compared to the promotion, that is a 17% reduction. 5-hour session limits were not part of the promotion. Anthropic then raised 5-hour limits on the same plans on September 22, 2026, with the Claude Opus 5.5 launch, and gave subscribers a limit reset they can save and use later.
Does the free Claude.ai plan include Claude Code?
No. Claude Code requires a Pro, Max, Team, Enterprise, or Console (API) account. Claude Pro is $17/mo billed annually ($200 up front) or $20/mo monthly and includes Claude Code. Codex is different: OpenAI lists Codex on ChatGPT Free ($0) and Go ($8/mo) for quick and lightweight tasks, with fuller use on Plus, Pro, Business, Edu, and Enterprise.
How many messages do you get with Codex per 5-hour window?
It depends on the model. ChatGPT Plus: GPT-6 Astra 5-45, GPT-6 Sol 15-150, GPT-5.6 Terra 25-200, GPT-5.5 15-80, GPT-6 Luna 350-3,000. Pro $100 (5x): Astra 25-225, Sol 70-700. Pro $200 (20x): Astra 100-900, Sol 300-3,000. GPT-5.5 leaves Codex on October 14, 2026. GPT-5.6 usage averages 5-30 credits per message, and Plus or Pro users who hit limits can buy additional credits.
What does Claude Code cost per month compared to Codex?
Entry tier is $20/mo for both. Claude Max is $100 (5x) or $200 (20x). ChatGPT Pro is $100 (5x) or $200 (20x, new sign-ups paused). Team plans: Claude Team standard seats are $20/seat/mo annual ($25 monthly), and ChatGPT Business is $20/user/mo annual ($25 monthly). Both also support API-key billing per token; Codex has no cloud features in API-key mode.
Is Codex better than Claude Code?
On price and limits it depends on the model you run. GPT-6 Astra burns Codex's allowance 3x faster than GPT-6 Sol on the same plan. On capability, this page does not rank them; the Codex vs Claude Code comparison covers benchmarks, subagents, and failure modes. As of September 22, 2026, Claude Code v2.1.280 defaults to Claude Opus 5.5 on every subscription plan, including Pro and Team Standard, which previously defaulted to Sonnet. Codex CLI 0.156.0 is current, with GPT-6 Sol and Luna newly available.
Source for the Hacker News figure: joshstrange on HN, September 22, 2026. For benchmarks and architecture, read the Codex vs Claude Code comparison.
Run DeepSeek and Open Models Without a Usage Window
Morph serves DeepSeek with full 16-bit activations and codegen-tuned speculative decoding. morph-dsv4flash is $0.141953125/1M input and $0.399625/1M output, billed per token with no 5-hour or weekly cap. WarpGrep, the agentic code search model, gives any coding agent better context for fewer tokens.
Sources
- OpenAI Codex pricing and limits
- Codex CLI releases (0.156.0)
- OpenAI Codex: install and surfaces
- TechCrunch: OpenAI puts Pro subscriptions on hold (Sept 10, 2026)
- TechCrunch: GPT-6 Sol and Luna (Sept 22, 2026)
- Claude pricing (Pro, Max, Team, API)
- Anthropic: What is the Max plan?
- Anthropic: Claude Code weekly limits promotion and Sept 14 change
- BleepingComputer: Claude Code weekly limits cut 17% vs promotion
- Claude Code releases (v2.1.280)
- Anthropic: Claude Opus 5.5 (Sept 22, 2026, 5-hour limit increase)
- Anthropic: What is a limit reset?
- OpenAI Codex models (GPT-6 Sol, Luna; GPT-5.5 retirement)
- Claude Code: setup and requirements