Limits comparison

Claude Code vs OpenAI Codex

Plans, usage limits, what happens at the limit, model scores, real build spend and use cases, side by side.

Data checked · every row links its source

The verdict, from the data

Both are coding agents in our grouping; decide on price and limits.

Better for you if…

Claude Code

  • Its best Claude model scores higher on Web apps, Agentic coding, UI design, Android.

Better for you if…

OpenAI Codex

No clear data edge in what we track.

It depends

  • Entry price: OpenAI Codex has no individual monthly price in our data, so we can't compare entry cost.
  • Usage per dollar: the two tools meter usage in different units, so we don't rank them on it.
  • Reported build spend: Claude Code median $100 (n=9), OpenAI Codex median $200 (n=3).

Each line is computed from our plan, score and cost data. It isn't a quality ranking. Data checked .

Latest models

Claude Code · latest Claude modelsClaude Haiku 5.5 Oct 7, 2026 · Claude Sonnet 5.5 Sep 28, 2026 · Claude Opus 5.5 Sep 22, 2026 · Claude Fable 5.1
OpenAI Codex · latest GPT modelsGPT-6.1 Sol Sep 29, 2026 · GPT-6 Luna Sep 22, 2026 · GPT-6 Astra Sep 3, 2026

Release dates from the vendors' official pages, checked Oct 17, 2026. Older models are hidden from the score charts below.

Price

Claude CodeOpenAI Codex

Cheapest paid plan
$20/mo
Not listed
Top individual plan
$200/mo
Not listed

Plans and limits

Claude Code

Claude Code plans
PlanPriceIncludedResetsAt the limit
Pro (Claude)$20/mo monthlyClaude Code included; usage limits, no fixed message countRolling 5-hour session window + weekly limits (weekly size not published)Usage-billed
Max 5x$100/mo5x Pro per-session usage; $100/mo API creditsSession limit resets every 5 hours; weekly limit across all models resets at a fixed weekly time assigned to your accountUsage-billed
Max 20x$200/mo20x Pro per-session usage; $200/mo API creditsSession limit resets every 5 hours; weekly limit across all models resets at a fixed weekly time assigned to your accountUsage-billed
Team Standard$25/seat/moClaude Code included with every seat; shared pool with ClaudeunverifiedNot confirmed
Team Premium$125/seat/moClaude Code included with every seat; shared pool with ClaudeunverifiedNot confirmed
Enterprise$20/seat/mo + API-rate usageClaude Code included with every seat; shared pool with ClaudeunverifiedNot confirmed

Sources: claude.com · support.claude.com · support.claude.com · checked 2026-10-16

OpenAI Codex

OpenAI Codex plans
PlanPriceIncludedResetsAt the limit
API key partialPay per use at OpenAI API pricingCodex CLI, SDK, IDE extension; no cloud features (GitHub review, Slack); models follow API key accessn/a (usage-based)Usage-billed

Sources: developers.openai.com · help.openai.com · checked 2026-10-16

Model scores by category

Best non-preliminary score from each tool's current model lineup (newest generation only) on one board per category. Older models are left out. Bars run from the lowest to the highest score on that board. These are third-party model scores, not tool scores.

Claude CodeOpenAI Codex

Web apps
1807 · claude-opus-5.5-max
1774 · gpt-6-astra-max
Agentic coding
64.85 · Opus 5.5 (max)
58.18 · GPT-6 Astra (max)
UI design
1389 · claude-opus-5-5
1383 · gpt-6-astra
Android
32.7 · Claude Opus 5.5
28 · GPT 6 Astra

LMArena WebDev Arena · Terminal-Bench 4.0 · Design Arena (All categories) · Android Bench 2.0 (long-horizon, 30 tasks)

What builders say they spent

Claude Code$50–$130median $100 · n = 9 · paid
OpenAI Codex$111.51–$200median $200 · n = 3 · paid

Estimate from user reports, p25–p75. How costs are counted →

Use cases

@froessell
Claude Code
Three.js arena-shooter web game launched on CrazyGames
HamadMalikKhan
Claude Code
Gave Claude Code a baseline website, ad accounts and a budget; the site is personalized per visitor with DeepSeek
yureki_lab
Claude Code
Audited and made 140 Celery background jobs idempotent after a double-charge incident
@joaopaulots
OpenAI Codex
A Snew landing page update: "For the first time in my life, I built something directly using Codex only."
Paras
OpenAI Codex
12 builds: a website about itself and a 3D Three.js Colosseum, each at all 6 reasoning efforts (low to ultra), about 7.5h of agent time. Max had the best code in a blind review
@kevincodex
OpenAI Codex
A music video built "using codex alone powered by GPT-6.1-Sol xhigh … cost is just $1.16"

All use cases →

FAQ

Is Claude Code or OpenAI Codex cheaper?

Our data lacks an individual monthly price for OpenAI Codex, so we can't say. See the plan tables for what is listed.

Does Claude Code or OpenAI Codex have a free plan?

Claude Code: no free plan listed. OpenAI Codex: no free plan listed.

What happens when you hit the limit on Claude Code vs OpenAI Codex?

Claude Code Pro (Claude): Usage-billed. OpenAI Codex API key: Usage-billed.

Which scores higher on coding benchmarks, Claude Code or OpenAI Codex?

LMArena WebDev Arena: claude-opus-5.5-max 1807 vs gpt-6-astra-max 1774; Terminal-Bench 4.0: Opus 5.5 (max) 64.85 vs GPT-6 Astra (max) 58.18; Design Arena (All categories): claude-opus-5-5 1389 vs gpt-6-astra 1383; Android Bench 2.0 (long-horizon, 30 tasks): Claude Opus 5.5 32.7 vs GPT 6 Astra 28. These are model scores from third-party boards, not tool scores.

More comparisons

The weekly diff

What changed in AI coding plans: prices, limits, resets. Email signups open soon. Until then, updates go out on X.

Follow @QuietlyByG on X RSS feed

No email form yet, so we don't collect any email addresses.