The comparison at a glance
Alibaba currently lists three Token Plan Individual tiers: Lite, Standard and Pro. At verification, the official page advertises promotional entry prices of $6, $18 and $68 per month respectively. Promotions can change, so checkout is authoritative for the price and terms available to a buyer.
All three tiers are intended for one person's interactive use in supported coding and agent tools. Paying for a higher tier increases the documented seven-day Credit limit and recommended number of concurrent agents. It does not turn the subscription into a general-purpose production API.
Published tier limits
Alibaba describes Standard as four times Lite usage and Pro as sixteen times Lite usage. Those ratios match the published seven-day quotas exactly. They should not be translated into a fixed number of tasks because each request's Credit deduction depends on its model, token volume, thinking mode and tool calls.
- Lite - promotional entry price $6 per month; 2,500 Credits per seven days; one to two recommended concurrent agents.
- Standard - promotional entry price $18 per month; 10,000 Credits per seven days; three to four recommended concurrent agents.
- Pro - promotional entry price $68 per month; 40,000 Credits per seven days; six to eight recommended concurrent agents.
Lite: a lower-cost way to learn the workflow
Lite provides the smallest seven-day allowance. It is the most conservative choice for someone configuring one supported tool, testing several included models and doing occasional interactive work. Its published one-to-two-agent guidance can also suit a developer who keeps only one primary coding assistant active.
The trade-off is that agentic work can consume several calls and tools for one visible task. Long reasoning sessions, repeated web-search calls or media generation may use more of the 2,500-Credit seven-day limit than a simple prompt suggests. Lite is therefore easier to assess when the user is willing to review consumption in the subscription console and narrow tasks.
Standard: more room for regular individual use
Standard increases the seven-day allowance to 10,000 Credits. Alibaba publishes a three-to-four concurrent-agent range for this middle tier. The larger documented allowance provides more flexibility for frequent coding assistance or switching among text, visual-understanding and creative models.
Recommended does not mean universally optimal. Someone whose tasks regularly hit Standard's seven-day ceiling may need Pro, but that conclusion should come from observed Credit history rather than an assumed prompt count.
Pro: higher seven-day capacity and concurrency
Pro raises the documented seven-day limit to 40,000 Credits, with guidance for six to eight concurrent agents. It is the Individual tier positioned for heavier interactive usage, multiple simultaneous agent sessions or workloads that combine model reasoning with Harness tools and multimodal generation.
Reaching the seven-day limit pauses plan calls unless the user applies an available Reset Usage Limit benefit or purchases an eligible Credit Pack, and Pro remains a single-user subscription. The higher concurrency figure is guidance rather than a performance assurance; Alibaba notes that peak periods can cause queuing. Pro should not be used as a workaround for a team, backend or scheduled automation.
Choose by measured workload, not headline capacity
A useful selection method is to identify the tools and model IDs you intend to use, estimate how concentrated your sessions are, then monitor actual Credit deductions. Short, efficient text tasks and long, tool-heavy agent loops do not consume alike. Image and video models also use separate integration patterns and can have materially different Credit deductions.
If Lite proves too restrictive, compare the current upgrade, Reset Usage Limit and Credit Pack options shown by Alibaba. Promotional prices and purchase terms can change, so checkout and the subscription console are authoritative. A lower entry tier can still be a reasonable starting point when requirements are uncertain and the buyer plans to measure real Credit use.