A decision for a beginner, not a spec sheet. Scores are our weights. Prices are the public plans, and anything marked verify on the tool page should be re-checked before you pay.
Pick Amp if you want a repo you want an agent to work on in parallel, including while your laptop is closed.
Pick OpenAI Codex if you already pay for ChatGPT and want the coding agent in that allowance.
A coding agent you can run free on your own machine, or in Amp's cloud.
Cheapest verified paid plan: Megawatt · $20/mo
A coding agent included with ChatGPT, not a separate app subscription.
Cheapest verified paid plan: Plus · $20/mo
Web app or SaaS
Mobile app
Game
Website or landing page
Automation or script
No code
I can follow a change
I already develop
Last verified 2026-10-19
Individual
Free
VerifiedAll product features. No monthly fee. Pay-as-you-go orbs, or your own runners at no Amp charge. Linked ChatGPT, Claude, and other subscriptions, and bring-your-own-key, with no Amp token fees or limits. No included orb minutes and no included Amp credits.
The Hobby card is on ampcode.com/pricing. Standard orb rates, if you do use cloud machines, are on the orb sizes page and start at $0.08 an hour for a1.tiny.
Individual
$20/mo
Verified45,000 minutes of orb time and $20 of Amp credits, then more credits at provider list price. Unlimited public and private repositories, and your own portal domains. Soft hosting limits on the card are 1 GB of repository storage and 10 GB of monthly transfer.
This is the Individual price shown on the default pricing tab. Included usage does not roll over. The $20 of credits is the amount the pricing card prints for this tier.
Individual
$10/mo
VerifiedThe same Megawatt allowance, at $10 a month for students and teachers. The pricing card shows $10 of included Amp credits when the education discount is applied.
The pricing FAQ says students and teachers can get Megawatt for $10 per month. Left out of budget matching so a general under-$10 question cannot be won by a discount most readers cannot buy.
Individual
$200/mo
Verified480,000 minutes of orb time, which is the Gigawatt tab's minute figure, and $200 of Amp credits. The underlying allowance is 1,000 hours of a1.xxlarge. a1.3xlarge orbs are limited to Gigawatt and Enterprise. Soft hosting limits are 10 GB of repository storage and 100 GB of monthly transfer.
The $200 price is the Gigawatt tier on ampcode.com/pricing. The default tab is Megawatt, so the $200 card is the other Individual tab. A note on the orb details says 10× limits for Gigawatt.
Last verified 2026-10-19
Individual
Free
VerifiedLimited trial access to Codex and limited access to GPT-5.6. Codex pricing docs also list GPT-6 Luna at Standard speed in the desktop app, subject to rollout. No published prompt count.
Confirmed on the Codex pricing page as a limited trial. GPT-6 Luna on Free is listed as subject to rollout, so it is not treated as a guarantee.
Individual
Confirm on chatgpt.com/pricing
VerifyLimited Codex, above Free. Codex pricing docs at learn.chatgpt.com/docs/pricing list Go at $8/mo. chatgpt.com/pricing still did not print a dollar amount on 2026-10-19.
Left out of budget matching. The $8 figure is on a Codex pricing doc, and the live ChatGPT pricing page still does not show it.
Individual
$20/mo
VerifiedExpanded Codex for a few focused sessions a week. Codex app on macOS and Windows. Local and cloud usage share the allowance. Weekly limits may apply. GPT-6 Astra access is limited, with credits for more.
Plus is $20/mo on OpenAI's help center. Annual billing is not offered.
Individual
$100/mo
VerifiedHigher Codex allowance for longer daily sessions. GPT-6 Astra is included in the allowance. Ultrafast is not included.
From OpenAI's Pro tiers help article.
Individual
$200/mo
VerifiedMore included usage than Pro 100. Ultrafast is not included. New subscriptions may have a lower allowance than older Pro 200 accounts.
OpenAI says the monthly price remains $200.
Individual
$500/mo
VerifiedHighest included Pro usage. Ultrafast for GPT-6 Astra and GPT-6.1 Sol is included and draws the included allowance at 8× the Standard rate for that model.
Ultrafast is only on this Pro tier. The 8× rate is from OpenAI's Pro tiers help article.
Included Megawatt and Gigawatt usage resets at the end of the monthly billing period and does not roll over. Purchased credits expire 12 months after you buy them. A paused orb costs nothing. Amp pauses an idle orb 5 minutes after the agent stops, or 20 minutes after your last interaction, whichever is later.
On Hobby, use your own runner if the bill you are avoiding is orb time. Model limits on a linked ChatGPT or Claude subscription are that vendor's cap, not an Amp cap.
Plan allowance, with weekly limits that may also apply. Local messages and cloud chats share it. OpenAI does not publish one universal reset clock.
Switch to a lower-cost model. OpenAI says that stretches the same allowance.
Amp Megawatt and OpenAI Codex Plus are both listed at $20/mo.
Amp is for a repo you want an agent to work on in parallel, including while your laptop is closed, people who already pay for ChatGPT or Claude and want to use that subscription inside an agent, and developers who can review agent output and do not need a hosted app builder.
OpenAI Codex is for developers who already pay for ChatGPT and want the coding agent in that allowance, web apps and automations with local and cloud tasks, and a few focused coding sessions a week on Plus.
Amp scores 1 out of 5 for someone who does not want to read code. OpenAI Codex scores 1 out of 5. The scores are the same. You are starting threads against a project and reading what the agent did. This is not a no-code builder. You are still in a coding agent. This is a poor pick if you never want to see a file.
Amp: Free to Amp if you bring your own model subscription or key and run on your own machine. Cloud orbs are pay-as-you-go. Hobby includes no orb minutes and no Amp credits. OpenAI Codex: Free includes a limited Codex trial. OpenAI does not publish a fixed prompt count. It is enough to see the product, not to build through the month.