On the morning of November 16, the phrase 红果AI漫剧热播榜排名TOP10 (Hongguo AI comic-drama hot-play Top 10) sat at No. 12 on Weibo's hot search. Hongguo is ByteDance's free, ad-supported short-drama app, and AI comic drama, a fully generated animation-style show assembled from image generation, image-to-video models, synthetic voices and editing [1], now makes up 95% of the new titles it adds [2]. The audience side of this trend is real and measured. The money side reads differently. The reporting that surrounds this chart shows chart position and creator income now moving on separate tracks, with the per-play payout that funds ordinary creators down roughly 90% in a year [3].

The hot-search phrase does not point at one list. The 中华网 (china.com) news feed republished a valid-plays board (红果霸屏榜) on November 16 that puts 《全球死考》 first at 198 million valid plays, 《穿越八零,我把娘俩宠成宝》 second at 90.1 million, 《满院亲戚全是上古大妖第七季》 third at 89.8 million, 《凡人百世书第九季》 fourth at 86.9 million and 《车粉人狠·公路求生》 fifth at 77.1 million, with every top-ten title above 75 million [2]. Tencent News published a heat-value (热度值) list on November 15 with a floor of 55.1 million and 《退役归来,这个保镖不好惹》 first at 69.3 million [4]. Two and a half weeks earlier, the content account 有料 ran the same heat metric with 《持械入宋:第十季》 first at 71.0 million and 《退役归来》 down in tenth place at 57.0 million [5]. None of those five named titles appears on either heat list, so the phrase is aggregating two different metrics that produce two different Top 10s. The next step for anyone following this trend is to read each Top 10 as a dated snapshot with a metric label, because the live platform chart could not be retrieved on the observation day and every number here belongs to the outlet that published it.

The audience the chart measures

The scale behind rank 12 shows up in published reports. DataEye-ADX counted 221,900 new AI dramas in the first half of 2026, roughly one per minute, and only 1,055 of them passed 100 million plays, a 0.47% breakout rate [3]. The same dataset puts 5.8% past 10 million plays, leaving more than 90% that never break out [6]. The market around that supply reached RMB 22 billion in January to May, is expected to hit RMB 40 billion for the full year on 138% growth, and counts more than 600 million users and 23,400 registered companies, with 515.7 billion new plays logged on Douyin-native AI dramas in the half [7].

Hongguo concentrates that attention. QuestMobile's monitoring records 168 million daily active users in July 2026, up 107% year over year, more than Tencent Video, iQiyi, Youku and MangoTV combined at about 155 million [8]. The counting method matters, because QuestMobile counts a day as active if the app was opened at all, and Hongguo's sign-in and coin incentives mean some of those opens are reward-driven rather than viewing [8]. Engagement holds up regardless, at 125 minutes per user per day as of February 2026 [8]. On the platform itself, 95% of new titles are already AI comic dramas [2], and eight of the ten entries on Hongguo's September 17 overall hot list were AI-made [9]. That is the chart face of the market. What the attention pays is a separate table.

What a chart slot pays

The supply numbers raise the question the hot search never states, which is who gets paid and how much. DataEye's figures put the industry's ad-share payout to creators at RMB 30 to 100 per 10,000 plays in the second half of 2025, falling to RMB 5 to 10 in 2026, a drop of nearly 90% [3]. The 中华网 repost carries a wider RMB 5 to 30 per 10,000 valid plays with no stated method or date [2]. Both ranges are outlet-reported and the denominators differ slightly, valid plays versus plays, so the two figures travel with their attributions.

At the No. 1 title's 198 million valid plays, the arithmetic gives RMB 99,000 to 198,000 on the DataEye range and up to about RMB 594,000 on the wider one. This is a labeled hypothetical calculation, not a reported payout. The source outlet itself estimates the likely take at roughly RMB 110,000 to 190,000 [2], and headline titles sit outside the ordinary per-play rate on platform-set guarantee and bonus terms [10]. Measured against a 50-million-play break-even line, fewer than 1.3% of AI dramas recoup their costs [3]. About 90% of AI short-drama companies are loss-making, the count of active production studios fell from 1,216 to 698 in the first quarter of 2026, and one screenwriter who uploaded 11 series of 16 episodes each collected a total of a few hundred RMB [3].

Costs moved the other way. Compute per episode rose from RMB 17 to RMB 80–100, per-minute costs from RMB 100 to RMB 200–300, and premium production runs at RMB 800–1,500 per minute, about RMB 100,000 for a 100-minute title [3]. Paid traffic is the third cost line, with creators reserving budget for the ad bidding that buys visibility while crowding keeps pushing bid prices up [3]. Seedance-based mid-tier output at RMB 800–1,200 per minute still comes to roughly one fifth of live-action cost [9]. Cheap production no longer guarantees profit, because payouts fell while model prices, compute and paid traffic all kept rising [3].

The tip of the pyramid still pays. 《万妖图录传》 opened its twelfth season on September 5 at 137 million heat, a platform record, and an unofficial estimate cited by 第一财经 (Yicai) puts season costs near RMB 100,000 and eleven-season payouts near RMB 20 million, a figure the report itself marks as unconfirmed [3]. In October 2025, two studios passed RMB 10 million in monthly payouts and 《盛夏芬德拉》 passed RMB 16 million within about six weeks of release [9]. On this evidence the market's income concentrated at the top while the base payout thinned, and chart rank measures where a title sits in that distribution.

September reset the income rules

The split between visibility and income has a dated cause. The NRTA's 《微短剧发展管理办法》 (Order No. 16) was adopted on July 27, published on July 30 and took effect on September 1, defining micro dramas as series with single episodes shorter than 20 minutes, sorting them into three tiers with filing and license requirements, and allowing fines up to RMB 100,000 for deviating from filed content [11]. Outlets quoting the order report that AI-generated dramas must carry a prominent label on every episode and that platforms may not use algorithms inducing addiction or overspending [9][3]. On September 4 the regulator required review before broadcast for everything uploaded [8].

Platform policy moved in the same days, as BT财经 reported it. The S+ AI-script guarantee fell from up to RMB 50,000 to RMB 10,000, 8,689 violating AI dramas had been removed by August, and the 2026 real-drama floor budget exceeds RMB 1.5 billion with a top single-title bonus of RMB 1.5 million [9]. The seven platforms courting AI comic-drama creators all require AI labels and filing, squeeze pure-pipeline AI output and encourage human-plus-AI fusion, with iQiyi paying a 100% share on exclusive premieres up to an S-level cap of RMB 1 million per title and Kuaishou investing up to RMB 2 million in a single S-level comic drama [10]. Tool prices rose alongside the rules, with Seedance 2.5 charging more while creators keep paying for its consistency advantage [3], and commissioned-production prices for overseas work sliding to RMB 600–800 per minute against RMB 3,000 domestic quotes last year [9].

The money now sits at the IP and platform layers. Yicai lists 阅文 Group's short-drama and AI comic-drama revenue at over RMB 430 million, up 2.3 times year over year, 昆仑万维's DramaWave AI short-drama platform above US$65 million in monthly flow at the end of June, with over 90% of its July new titles AI-generated, and 中文在线's short-drama and IP-derivative revenue up 108.72% to RMB 411 million [3]. On the platform side, MangoTV's parent saw first-half net profit fall 73.58% [3], iQiyi posted a second-quarter net loss of RMB 290 million [8], and Tencent Video's membership fell 3 million year over year, a decline the report links to short-drama impact [8]. The loop feeding the chart runs Fanqie Novel IP through Douyin traffic into Hongguo monetization on Volcano Engine compute, with over 500,000 IP adaptation rights open [9], and one AI screenwriter's complaint to BT财经 captures the position most creators occupy, buy its compute, use its tools, publish on its platform [9]. Wang Ke of 智象未来 gave Yicai the one-line version, "AI has solved low-cost production; it has not yet solved low-cost visibility" [3]. Every audience figure on record predates the September rules, and what the rules repriced is who gets paid for feeding that audience.

What the next list can carry

A single position on these lists means little even within one metric, because the heat ranking churns weekly. 《退役归来,这个保镖不好惹》 climbed from tenth to first and 《持械入宋:第十季》 slipped from first to second between late October and mid November [4][5]. For anyone tracking this market, two published series matter more than rank, the per-10,000-plays payout line and the breakout rate, both carried in DataEye's half-year reports [3][6], plus the date and metric label that decide what any future Top 10 can support.

On the production side, the tool market has split into modular generation tools and full-chain platforms, with Kling 4.0, Seedance and Hailuo H3 among the modular options, LibTV and Tago Movie chaining clips into complete episodes, and free credits that zero out trial costs [1], while the documented failure point remains consistency, the shot-to-shot face swaps that still break fully generated episodes [1]. The breakout differentiator in the data is IP, since series hits mostly adapt web novels and pure originals rarely sustain a series [6]. Rank 12 marks the audience's arrival. The income question is answered by payout tables, filing labels and IP ownership, and chart position alone settles none of it.