On November 9, Alibaba Cloud raised subscription quotas across the board for the personal edition of its Token Plan. Monthly fees for the Lite, Essential, and Standard tiers remain unchanged, while monthly quotas increased by 74% to 78% respectively. The Pro tier is even more straightforward: its quota jumps from 180,000 to 200,000 Credits, and the monthly fee drops from 499 yuan to 312 yuan.

For the cheapest Lite tier, the monthly quota rises from 11,500 Credits to 20,000, with the monthly fee still at 39 yuan. Essential and Standard follow suit: the former goes from 25,500 to 45,000 at 79 yuan per month, while the latter goes from 45,000 to 80,000 at 139 yuan per month. All three tiers now offer more for the same price. As for Pro, its quota increases by only 20,000, yet the monthly fee drops directly from 499 yuan to 312 yuan—a 38% decrease.

Calculating monthly fee divided by monthly quota, the unit quota cost for all four tiers—Lite, Essential, Standard, and Pro—has dropped by over 40%. Take the cheapest Lite tier as an example: before the adjustment, 39 yuan bought 11,500 Credits at about 0.0034 yuan per unit; after the adjustment, the same 39 yuan buys 20,000 Credits at about 0.002 yuan per unit—nearly cut by 40%.

This adjustment isn't just a sweetener for new customers—existing users are included too. Starting November 9, the new quotas for all four tiers take effect gradually, and the total quota in the current subscription cycle is topped up according to the new standard. Existing Pro users keep the same price for this cycle, with the new price starting from the next cycle. Most platforms offer new discounts only to new registrants, but Alibaba Cloud tops up existing users as well—the signal this sends is different from pure customer acquisition.

Token Plan is a large model subscription service launched by Alibaba Cloud Bailian. It can connect to AI programming and agent tools such as Qwen Office, Qwen APP, Qoder, Qwen Code, and Claude Code, using Credits to uniformly offset the costs of different model calls and Agent Harness tools, covering both text and multimodal models. Using these quotas to run image or video generation models also deducts from Credits.

For individual developers, the most practical aspect of this subscription set is that it simplifies billing. In the past, calling different models required separate API applications and separate bill tracking; now a single Credits pool covers text, code, and multimodal, so a single top-up at the start of the month allows open experimentation. With more quota, those who benefit most are people who write code daily with Claude Code or Qwen Code—revisions and debugging no longer require counting every call.

It's worth noting that this quota increase coincides with the Double 11 promotion. Alibaba Cloud has slashed Token Plan to a limited-time 50% off, with new customers getting Lite as low as 19.5 yuan and Essential as low as 39.5 yuan for the first month. More quota at no extra cost. Combined with the first-month half-price, the entry barrier for individual developers this month has been pushed very low.

Additionally, beyond the personal edition, Alibaba Cloud's team edition has long had lower pricing. The cheapest standard seat is 150 yuan per month for 25,000 Credits; the advanced seat at 550 yuan gets 100,000; and the top premium seat at 1,398 yuan provides 250,000. Both personal and team lines are moving downward together—the whole thing looks more like a round of concessions across the entire AI subscription system.

One thing to note: personal edition quotas reset on a monthly subscription cycle, and unused Credits do not roll over—they are cleared to zero at the end of the cycle. So this quota increase benefits those who dare to use it and use it frequently within the month, not those who stockpile it unused.

Looking back at Alibaba Cloud itself, it has been cutting prices throughout 2026. The Bailian platform has repeatedly lowered model API prices this year. In January, qwen3-max was the first to drop; in August, Qwen3-VL-Rerank's text and multimodal inputs were both reduced; and by September, the speech series models followed suit. The Token Plan quota increase is a continuation of this price-cutting line, except the battlefield has shifted from pay-per-use to subscription packages.

The price war among domestic large models has stratified by 2026. According to multiple price reviews during the year, DeepSeek permanently cut output prices by 75%, while Zhipu GLM bucked the trend with an 80%+ increase—yet call volume rose fourfold. Domestically, inference costs for domestic models have dropped by over 90% in two years, giving Alibaba Cloud room to continue increasing quotas on the subscription side.

It chose the path of increasing quotas, and by including existing users, it aims to keep developers within its own Agent ecosystem. Qoder, Qwen Code, and Claude Code are all integrated into Token Plan; with more quota, users are more willing to try multiple rounds of trial and error when writing code and tuning Agents in these tools.

For domestic individual developers, the most tangible aspect of this adjustment is that unit cost has dropped by 40%, and existing users can benefit too. Domestically quotas are increasing while overseas prices are rising—this scissors gap is rare in the 2026 AI tool market.

It is understood that in 2026, overseas AI programming tools have collectively raised prices. Cursor was repriced after being acquired in October, eliminating the $20 Pro tier; Claude Code pushed heavy users to the $100–200 tier; Windsurf went from $15 to $20.

Some overseas reviews directly call 2026 "the end of the $20 AI programming era." The logic behind this is that frontier model inference costs cannot be contained, and vendors are no longer subsidizing customer acquisition as they did in 2024–2025.