Claude Code vs Codex: which one's apps make more money?

• $14,500median monthly revenue, Claude Code (22 cases)
• $13,100median monthly revenue, Codex (8 cases)
• 14 of 36Claude Code cases verified
• 3 of 15Codex cases verified

Which one makes more money?

In our October 2026 data the medians are close, with Claude Code slightly ahead: $14,500 a month across 22 cases with a monthly US dollar figure, vs $13,100 for Codex (8 cases). Verified Claude Code cases with a monthly US dollar figure: 12, median $2,280. Verified Codex cases with a monthly US dollar figure: 2, too few for a median.

These are small samples (22 and 8 monthly figures), so one or two new cases can move a median a lot. Most figures are the founder's own claim: 14 of 36 (39%) Claude Code cases and 3 of 15 (20%) Codex cases are verified through the payment provider. And the tool doesn't cause the revenue: the product, the market and how the founder gets customers matter far more. Verified medians run lower partly because verified cases come from public revenue profiles, which list many small apps, while self-reported ones mostly come from posts and interviews that favor big wins.

Side by side

A case counts for a tool when the founder or source names it; many name several tools. Sources call it “Codex” or “OpenAI Codex”, so both names count as one tool here, as in the report. Figures that aren’t a monthly US dollar amount (totals, ARR, other currencies) count in “Apps naming the tool” but not in the medians.

Example cases

Claude Code: highest monthly figures

1. Dropmagic: $120K MRR (founder on X, October 2026); over 45K MRR within the first four months Self-reported
2. Kleo: $61,910 MRR, about 90 days after relaunch Screenshot
3. Erly: Over $50,000 per month, about four months after launch Screenshot
4. Laper: $46,509 MRR Verified
5. Vaultaire: $41,425 in gross App Store sales in August 2026 Self-reported

Verified Claude Code cases (14): Laper ($46,509 MRR), ClaudeKit ($279 all time), ZenMode ($1,729 MRR), SpacePlanner ($711 in the last 30 days), Aidelly ($655 MRR), GETebook.ai ($35,469 in the last 30 days), Frameliq ($631 MRR), VectoSolve ($2,695 in the last 30 days), SocialCrawl ($35,135 in the last 30 days), Tucket ($672 in the last 30 days), CNVS ($49,986 in all-time sales), Rewired ($1,864 MRR), SuperX ($22,281 MRR), Visualizee ($8,204 MRR).

All 36 cases: apps built with Claude Code.

Codex: highest monthly figures

1. Laper: $46,509 MRR Verified
2. Vaultaire: $41,425 in gross App Store sales in August 2026 Self-reported
3. Tarsi: About $20,000 in its first month Screenshot
4. NinjaPear: Over $15k in gross revenue in a month, about six months after a January 2026 launch Self-reported
5. Vugola: about $11,200 MRR Self-reported

Verified Codex cases (3): Laper ($46,509 MRR), Frameliq ($631 MRR), CNVS ($49,986 in all-time sales).

All 15 cases: apps built with OpenAI Codex.

Apps built with both (10)

These founders name both Claude Code and Codex, so they count on both sides above.

• Laper: $46,509 MRR Verified
• Frameliq: $631 MRR Verified
• CNVS: $49,986 in all-time sales Verified
• Vaultaire: $41,425 in gross App Store sales in August 2026 Self-reported
• NinjaPear: Over $15k in gross revenue in a month, about six months after a January 2026 launch Self-reported
• BridgeMind: ARR: $250,772 Self-reported
• Clairvo: a million dollars in ARR Self-reported
• Profit AI: 30K MRR for the app Screenshot
• Coherence: around 85,000 in eight months Self-reported
• ContextBolt: $653 a month Self-reported

How we compared

• Data. Every published case in our report How much AI-built apps really make (October 2026), data as of Oct 22, 2026 (85 apps). Each case links to the founder’s own source.
• Figures. The headline number on each case page as the source states it, counted only if it is per month and in US dollars. Nothing is converted, annualized or estimated.
• Medians, not averages, shown only where at least 3 cases give a monthly US dollar figure.
• Survivorship bias. These are apps that made money and said so in public. They are not typical results for either tool and can’t show that one tool earns more.
• Reproducible. Same rules and dataset as the report: CSV · methodology. Free to cite with a link (CC BY 4.0).