AI Game Index beta

Updated October 28, 2026Data September 28 – October 27, 2026

Which AI agents and models make the games people actually play, judged by real players instead of benchmarks.

Based on 5,300+ games and 1.4M+ launches in the Playgama AI sandbox

The AI Game Index is a leaderboard that ranks AI coding agents, such as Claude Code, Codex, Cursor and Google Antigravity, and the large language models behind them, such as Claude Opus, Claude Sonnet, OpenAI Sol, OpenAI Astra, Gemini and Grok, by how real players treat the games they build. Unlike traditional benchmarks, which score code on set tasks, it is a behavioral evaluation: play time and likes are observed from the players of every game published to the Playgama AI sandbox, with no human judges. The agent and model behind each game are self-reported by its developer, and the ranking is updated daily.

AI coding agents compared

The agent is the tool that wrote the game: a coding agent such as Claude Code (Anthropic), Codex (OpenAI), Cursor or Google Antigravity, or a chat app such as ChatGPT or the Claude app on the web and Desktop. The developer names it when they publish.

Share of new games by agent

Claude Code
Codex
Cursor
Google Antigravity
OpenCode
Gemini CLI
Grok Build
WorkBuddy
ZCode
Claude app
ChatGPT
Other

Table data

Share of new games by agent, per day
Day	Claude Code	Codex	Cursor	Google Antigravity	OpenCode	Gemini CLI	Grok Build	WorkBuddy	ZCode	Claude app	ChatGPT	Other
Sep 28	39%	28%	1.3%	4.0%	–	–	1.3%	4.0%	–	6.7%	2.7%	13%
Sep 29	30%	34%	1.2%	–	–	–	2.4%	2.4%	–	16%	4.9%	8.5%
Sep 30	41%	21%	1.2%	4.7%	–	–	–	–	1.2%	20%	7.0%	4.7%
Oct 1	32%	12%	10%	3.8%	–	–	2.6%	1.3%	–	22%	7.7%	9.0%
Oct 2	52%	23%	–	1.4%	1.4%	1.4%	–	–	–	8.5%	5.6%	7.0%
Oct 3	45%	14%	9.1%	–	–	1.5%	–	–	1.5%	17%	3.0%	9.1%
Oct 4	40%	21%	6.2%	3.7%	4.9%	1.2%	1.2%	2.5%	–	12%	3.7%	3.7%
Oct 5	44%	21%	2.4%	3.5%	–	–	–	4.7%	–	15%	2.4%	7.1%
Oct 6	48%	20%	3.4%	2.3%	3.4%	–	–	5.7%	–	10%	2.3%	4.5%
Oct 7	42%	24%	2.0%	1.0%	2.9%	–	–	3.9%	–	16%	2.9%	5.9%
Oct 8	55%	16%	1.4%	2.1%	2.1%	–	1.4%	2.1%	0.7%	15%	2.1%	2.1%
Oct 9	54%	17%	4.2%	0.7%	1.4%	0.7%	–	–	0.7%	6.3%	2.8%	12%
Oct 10	56%	19%	0.8%	1.5%	0.8%	–	–	0.8%	0.8%	2.3%	5.3%	13%
Oct 11	54%	19%	1.5%	1.5%	5.2%	0.7%	–	3.0%	–	5.2%	2.2%	7.5%
Oct 12	58%	19%	–	1.3%	1.3%	–	–	2.0%	–	11%	3.3%	4.0%
Oct 13	48%	28%	–	2.0%	2.0%	1.3%	0.7%	2.0%	–	4.6%	3.3%	8.6%
Oct 14	56%	21%	1.3%	4.4%	–	0.6%	–	1.9%	–	6.3%	3.8%	4.4%
Oct 15	55%	20%	0.5%	2.9%	3.8%	0.5%	–	1.0%	–	6.7%	3.8%	5.7%
Oct 16	54%	16%	2.2%	7.5%	3.8%	0.5%	–	2.7%	1.6%	4.3%	2.2%	4.8%
Oct 17	57%	18%	1.2%	4.7%	2.9%	0.6%	0.6%	–	1.2%	4.1%	4.1%	5.8%
Oct 18	49%	15%	3.9%	3.9%	0.5%	–	–	0.5%	–	9.8%	4.9%	12%
Oct 19	51%	17%	2.4%	4.3%	0.5%	0.5%	–	–	–	10%	3.3%	11%
Oct 20	60%	15%	1.0%	5.4%	1.5%	0.5%	–	1.0%	0.5%	7.8%	1.0%	6.8%
Oct 21	48%	14%	2.2%	4.3%	2.2%	–	0.5%	–	0.5%	13%	5.9%	9.2%
Oct 22	57%	21%	1.7%	0.4%	0.4%	0.9%	–	–	–	9.1%	1.7%	7.8%
Oct 23	63%	8.8%	1.4%	0.9%	0.9%	0.9%	–	1.9%	1.4%	14%	1.4%	5.1%
Oct 24	48%	13%	1.0%	6.7%	0.5%	0.5%	–	2.6%	0.5%	16%	1.5%	8.8%
Oct 25	53%	14%	0.5%	2.1%	1.6%	0.5%	–	3.7%	0.5%	14%	4.2%	5.2%
Oct 26	50%	21%	1.5%	2.4%	0.5%	–	–	–	0.5%	13%	4.4%	6.3%
Oct 27	45%	14%	4.5%	–	1.0%	0.5%	1.5%	1.5%	–	21%	3.5%	7.5%

All time, by agent

Claude Code 51%
Codex 18%
Claude app 10%
Other 8.4%
Google Antigravity 3.1%
ChatGPT 3.1%
Cursor 2.1%
WorkBuddy 1.6%
OpenCode 1.4%
Gemini CLI 0.5%
ZCode 0.4%
Grok Build 0.3%

Table data

Share of games by agent, all time
Agent	Share
Claude Code (Anthropic)	51%
Codex (OpenAI)	18%
Claude app (Anthropic)	10%
OtherKimi Code Agent	8.4%
Google Antigravity	3.1%
ChatGPT (OpenAI)	3.1%
Cursor (Anysphere)	2.1%
WorkBuddy	1.6%
OpenCode	1.4%
Gemini CLI (Google)	0.5%
ZCode	0.4%
Grok Build (xAI)	0.3%

Play time by agent

How long a game holds its players: the average length of its plays, where a play is a session of 30 seconds or more. Each row spreads over the agent’s games with 20 plays or more, and the white mark is the median game, so one hit cannot carry a row.

OpenCode

4 min 21 s3 min 10 s – 6 min 46 s
Google Antigravity

4 min 19 s3 min 3 s – 6 min 35 s
Claude app

4 min 13 s3 min 5 s – 6 min 3 s
Codex

4 min 9 s2 min 57 s – 7 min 25 s
Claude Code

4 min 6 s3 min 3 s – 6 min 7 s
Cursor

4 min2 min 59 s – 5 min 11 s
WorkBuddy

4 min3 min 12 s – 6 min 3 s
ChatGPT

3 min 34 s2 min 43 s – 4 min 33 s

10th to 90th percentile
20th to 80th percentile
median game

Not enough data yet to rate Gemini CLI, Grok Build and ZCode. Other is not rated: it pools the games whose developer picked Other with every agent under 10 games.

Table data

Play time by agent
Agent	Median game	20th to 80th percentile	10th to 90th percentile
OpenCode	4 min 21 s	3 min 10 s – 6 min 46 s	2 min 33 s – 9 min 12 s
Google Antigravity	4 min 19 s	3 min 3 s – 6 min 35 s	2 min 53 s – 7 min 42 s
Claude app (Anthropic)	4 min 13 s	3 min 5 s – 6 min 3 s	2 min 44 s – 8 min 4 s
Codex (OpenAI)	4 min 9 s	2 min 57 s – 7 min 25 s	2 min 32 s – 9 min 33 s
Claude Code (Anthropic)	4 min 6 s	3 min 3 s – 6 min 7 s	2 min 41 s – 8 min 9 s
Cursor (Anysphere)	4 min	2 min 59 s – 5 min 11 s	2 min 41 s – 6 min 14 s
WorkBuddy	4 min	3 min 12 s – 6 min 3 s	2 min 57 s – 7 min 18 s
ChatGPT (OpenAI)	3 min 34 s	2 min 43 s – 4 min 33 s	2 min 14 s – 5 min 30 s

Likes per 1,000 plays by agent

How much players like a game: its likes for every 1,000 of its plays. Each row is the median of the agent’s games with 20 plays or more, so one hit cannot carry it.

Codex

59
OpenCode

54
ChatGPT

53
Cursor

41
Claude Code

40
Google Antigravity

34
Claude app

33
WorkBuddy

32

Not enough data yet to rate Gemini CLI, Grok Build and ZCode. Other is not rated: it pools the games whose developer picked Other with every agent under 10 games.

Table data

Likes per 1,000 plays by agent
Agent	Likes per 1,000 plays
Codex (OpenAI)	59
OpenCode	54
ChatGPT (OpenAI)	53
Cursor (Anysphere)	41
Claude Code (Anthropic)	40
Google Antigravity	34
Claude app (Anthropic)	33
WorkBuddy	32

AI models compared

The model is the large language model the agent ran on: Claude Opus, Sonnet or Fable (Anthropic), OpenAI Sol or Astra, Gemini (Google), Grok (xAI) and others. The form records the model, not its version. One agent can run on several models, so the two views do not line up one to one.

Share of new games by model

Claude Fable
Claude Opus
Claude Sonnet
OpenAI Astra
OpenAI Sol
OpenAI Luna
OpenAI Terra
Gemini Pro
Gemini Flash
Grok
DeepSeek
GLM
Other

Table data

Share of new games by model, per day
Day	Claude Fable	Claude Opus	Claude Sonnet	OpenAI Astra	OpenAI Sol	OpenAI Luna	OpenAI Terra	Gemini Pro	Gemini Flash	Grok	DeepSeek	GLM	Other
Sep 28	–	43%	7.1%	10%	11%	–	–	–	1.4%	1.4%	4.3%	–	21%
Sep 29	2.6%	47%	3.9%	17%	6.5%	–	–	–	–	3.9%	2.6%	–	17%
Sep 30	–	55%	3.6%	9.5%	11%	–	1.2%	–	6.0%	–	–	1.2%	13%
Oct 1	5.1%	46%	8.9%	8.9%	10%	–	–	1.3%	–	5.1%	–	–	15%
Oct 2	–	54%	7.7%	4.6%	11%	1.5%	–	1.5%	–	–	–	–	20%
Oct 3	–	57%	6.2%	3.1%	6.2%	3.1%	–	–	1.5%	–	–	1.5%	22%
Oct 4	1.3%	47%	5.1%	2.6%	13%	–	–	1.3%	3.8%	1.3%	–	1.3%	23%
Oct 5	2.7%	55%	8.0%	–	16%	–	–	2.7%	4.0%	1.3%	1.3%	–	9.3%
Oct 6	1.3%	51%	11%	2.6%	13%	–	–	1.3%	1.3%	1.3%	–	–	17%
Oct 7	–	51%	7.4%	3.2%	12%	–	1.1%	1.1%	–	–	2.1%	–	23%
Oct 8	0.7%	59%	12%	0.7%	7.2%	0.7%	–	–	0.7%	2.2%	2.9%	0.7%	12%
Oct 9	–	56%	11%	–	12%	0.8%	–	0.8%	0.8%	–	0.8%	0.8%	16%
Oct 10	4.1%	52%	5.7%	1.6%	15%	0.8%	0.8%	1.6%	–	0.8%	2.5%	0.8%	14%
Oct 11	3.2%	56%	6.3%	1.6%	16%	–	0.8%	0.8%	1.6%	–	1.6%	–	13%
Oct 12	3.5%	60%	8.5%	0.7%	18%	0.7%	–	–	0.7%	–	0.7%	–	7.0%
Oct 13	5.0%	50%	2.9%	2.9%	16%	0.7%	–	1.4%	1.4%	0.7%	3.6%	–	15%
Oct 14	4.3%	53%	11%	5.0%	13%	–	–	1.4%	2.2%	0.7%	–	–	10%
Oct 15	4.0%	60%	6.9%	2.9%	8.1%	0.6%	–	2.3%	1.2%	0.6%	1.7%	0.6%	11%
Oct 16	4.5%	59%	5.3%	2.3%	6.8%	0.8%	–	7.6%	2.3%	1.5%	5.3%	2.3%	2.3%
Oct 17	2.6%	67%	3.4%	2.6%	8.5%	–	–	4.3%	3.4%	2.6%	2.6%	0.9%	2.6%
Oct 18	1.4%	70%	6.4%	2.1%	7.1%	–	0.7%	5.0%	3.6%	–	1.4%	–	2.1%
Oct 19	3.9%	58%	14%	1.3%	7.1%	0.6%	–	5.8%	1.3%	1.3%	1.9%	0.6%	3.9%
Oct 20	3.4%	62%	9.4%	4.0%	7.4%	–	–	2.7%	4.0%	1.3%	2.0%	0.7%	3.4%
Oct 21	6.3%	53%	11%	–	12%	–	–	2.1%	4.2%	4.2%	3.5%	–	3.5%
Oct 22	9.6%	57%	13%	1.8%	11%	–	–	–	1.8%	1.8%	1.2%	–	3.0%
Oct 23	7.9%	68%	14%	0.7%	2.0%	–	–	0.7%	1.3%	0.7%	2.0%	0.7%	2.6%
Oct 24	5.3%	58%	7.3%	–	10%	–	–	8.7%	0.7%	5.3%	2.7%	1.3%	0.7%
Oct 25	6.4%	63%	11%	1.4%	6.4%	–	–	–	2.1%	–	3.5%	0.7%	5.0%
Oct 26	1.4%	65%	8.8%	3.4%	13%	0.7%	–	0.7%	1.4%	2.0%	0.7%	–	3.4%
Oct 27	0.7%	76%	6.8%	0.7%	6.1%	–	0.7%	0.7%	–	3.4%	–	0.7%	4.7%

All time, by model

Claude Opus 56%
OpenAI Sol 10%
Other 10%
Claude Sonnet 8.6%
Claude Fable 3.5%
OpenAI Astra 3.3%
Gemini Pro 2.0%
Gemini Flash 1.8%
DeepSeek 1.7%
Grok 1.5%
GLM 0.5%
OpenAI Luna 0.4%
OpenAI Terra 0.3%

Table data

Share of games by model, all time
Model	Share
Claude Opus (Anthropic)	56%
OpenAI Sol	10%
OtherClaude Haiku, Kimi, Qwen	10%
Claude Sonnet (Anthropic)	8.6%
Claude Fable (Anthropic)	3.5%
OpenAI Astra	3.3%
Gemini Pro (Google)	2.0%
Gemini Flash (Google)	1.8%
DeepSeek	1.7%
Grok (xAI)	1.5%
GLM (Zhipu AI)	0.5%
OpenAI Luna	0.4%
OpenAI Terra	0.3%

Play time by model

How long a game holds its players: the average length of its plays, where a play is a session of 30 seconds or more. Each row spreads over the model’s games with 20 plays or more, and the white mark is the median game, so one hit cannot carry a row.

OpenAI Astra

4 min 46 s3 min 21 s – 8 min 23 s
Gemini Pro

4 min 36 s3 min 17 s – 6 min 55 s
Grok

4 min 34 s3 min 32 s – 5 min 42 s
Claude Opus

4 min 14 s3 min 8 s – 6 min 21 s
OpenAI Sol

4 min 12 s3 min – 8 min 15 s
DeepSeek

3 min 55 s3 min 4 s – 5 min 4 s
Claude Sonnet

3 min 46 s2 min 43 s – 6 min 4 s
Claude Fable

3 min 41 s2 min 53 s – 4 min 58 s
Gemini Flash

3 min 20 s2 min 45 s – 4 min 26 s

10th to 90th percentile
20th to 80th percentile
median game

Not enough data yet to rate OpenAI Luna, OpenAI Terra and GLM. Other is not rated: it pools the games whose developer picked Other with every model under 10 games.

Table data

Play time by model
Model	Median game	20th to 80th percentile	10th to 90th percentile
OpenAI Astra	4 min 46 s	3 min 21 s – 8 min 23 s	2 min 52 s – 9 min 39 s
Gemini Pro (Google)	4 min 36 s	3 min 17 s – 6 min 55 s	2 min 56 s – 8 min 20 s
Grok (xAI)	4 min 34 s	3 min 32 s – 5 min 42 s	2 min 58 s – 6 min 51 s
Claude Opus (Anthropic)	4 min 14 s	3 min 8 s – 6 min 21 s	2 min 47 s – 8 min 21 s
OpenAI Sol	4 min 12 s	3 min – 8 min 15 s	2 min 33 s – 12 min 7 s
DeepSeek	3 min 55 s	3 min 4 s – 5 min 4 s	2 min 44 s – 6 min 3 s
Claude Sonnet (Anthropic)	3 min 46 s	2 min 43 s – 6 min 4 s	2 min 32 s – 7 min 40 s
Claude Fable (Anthropic)	3 min 41 s	2 min 53 s – 4 min 58 s	2 min 41 s – 8 min 22 s
Gemini Flash (Google)	3 min 20 s	2 min 45 s – 4 min 26 s	2 min 31 s – 7 min 5 s

Likes per 1,000 plays by model

How much players like a game: its likes for every 1,000 of its plays. Each row is the median of the model’s games with 20 plays or more, so one hit cannot carry it.

DeepSeek

80
OpenAI Astra

75
Grok

55
OpenAI Sol

53
Claude Sonnet

46
Claude Opus

38
Gemini Pro

34
Gemini Flash

34
Claude Fable

29

Not enough data yet to rate OpenAI Luna, OpenAI Terra and GLM. Other is not rated: it pools the games whose developer picked Other with every model under 10 games.

Table data

Likes per 1,000 plays by model
Model	Likes per 1,000 plays
DeepSeek	80
OpenAI Astra	75
Grok (xAI)	55
OpenAI Sol	53
Claude Sonnet (Anthropic)	46
Claude Opus (Anthropic)	38
Gemini Pro (Google)	34
Gemini Flash (Google)	34
Claude Fable (Anthropic)	29

How the AI Game Index is calculated

Where the games come from: Every game was published to the Playgama AI sandbox, where people ship web games they built with AI, by hand or through Playgama MCP straight from their agent. When they publish, the developer names the agent and the model they used. The answers are self-reported and not verified.
Share of new games: Each day’s new games, split by the agent or model their developer named. A game counts once, on the day it was first published, and keeps its place if it is later taken down.
Play time: How long a game holds its players: the all-time average length of its plays, where a play is a session of 30 seconds or more, as counted by Playgama Bridge, the SDK inside the game. A row shows the spread of that figure across its games: the median game and the 10th, 20th, 80th and 90th percentiles. The median, not the mean, so one hit game cannot carry a row.
Likes per 1,000 plays: Each game’s all-time likes on its sandbox page over its all-time plays, times 1,000. A row is the median game, and a game nobody liked counts as zero.
Minimum sample: A game is measured once it has 20 plays. An agent or model is rated once 15 of its games are measured; until then it is named under the chart as not having enough data yet.
What Other means: The games whose developer picked Other, and every agent or model named by fewer than 10 games; the tables list them under Other. Other counts in the shares but is not rated, since a figure for a mix of tools rates no one in particular.
The totals at the top: Every game live in the sandbox and all of their launches since the sandbox opened, as Playgama Bridge counts them, including the games that name no agent or model.
Updates: Daily. A day closes at 00:00 UTC and lands here about an hour later; the date of the last update is next to the title.

Limitations

Self-reported. The agent and model are the developer’s answer and are not checked. A developer who changes the answer moves the game, with its history, to the new row.
One platform. Only games published to the Playgama AI sandbox count. The index says nothing about games published elsewhere.
Uneven traffic. Some games draw far more players than others, from promotion or links shared outside Playgama. The median keeps one popular game from carrying a row, but a row with few games can still swing.
All-time figures. Play time and likes cover each game’s whole life, not the last week, so a newer agent or model has had less time to collect them.
No model versions. The form names the model, not the version: Claude Opus stands for whichever Opus the developer ran.

Frequently asked questions

Which AI is best for making games?

The AI Game Index answers with what players do rather than with a test: it ranks AI coding agents and models by how long people play the games they build and how many likes those games earn. The leader depends on the measure, since the agent that ships the most games is not always the one whose games hold players longest, so read the play time and likes rankings side by side.

Is Codex better than Claude Code for building games?

Compare their rows in the agent section: the share of new games, the median play time and the likes per 1,000 plays, measured on the same platform and the same players. A gap of a few seconds in play time sits well inside the spread of either agent’s games, which the percentile bars show.

What is the best AI model for vibe coding games?

The model section ranks the large language models behind published games, among them Claude Opus, Claude Sonnet, OpenAI Sol, OpenAI Astra, Gemini and Grok, by the play time and likes of their games. A model is rated once 15 of its games have 20 plays or more; until then it is named under the chart as not having enough data yet.

How is the AI Game Index calculated?

From observed player behaviour, with no human judges. Play time is the average length of a game’s plays, where a play is a session of 30 seconds or more; likes are counted per 1,000 plays; each agent or model gets the median of its games. The methodology lists every threshold.

Where do the agent and model labels come from?

From the developers: when they publish a game to the Playgama AI sandbox, they name the agent and the model they used. The labels are self-reported and not verified, and the form records the model, not its version.

How often is the index updated?

Every day. A day closes at 00:00 UTC and lands on the page about an hour later; the date of the last update is shown next to the title.

How can my game be in the index?

Publish it to the Playgama AI sandbox with Playgama MCP and name the agent and model you used. It counts in the shares from the day it is published, and in play time and likes once it has 20 plays.

What’s next

The index is in beta. More agents, models and measures are on the way as more games ship. Follow Playgama to hear when it grows.

Built a game with AI? Publish it with Playgama MCP and your agent joins the index.