
当 Token 从消耗品变为流通货币:360 送出 1 亿 Token 的背后,是整个算力交易逻辑的重塑
2026 年 7 月 28 日,360 集团创始人周鸿祎正式发布新一代企业智能体工作平台纳米Work,为中小企业 AI 规模化落地按下加速键。发布会重磅福利政策落地:平台将为首批入驻用户每人发放 1 亿 Token 专属试用额度,搭配一对一专业实训、全程陪跑落地服务,精准助力 1000 家中小微企业零门槛完成 AI 业务落地、实现经营效率升级。
很多人只看到“免费送 Token”的福利噱头,却忽略了这场发放背后的产业变革。以当前主流大模型算力定价折算,单人 1 亿 Token 额度等价于数万美元优质算力资源,360 此番批量发放数十亿级 Token,绝非简单的营销补贴,而是一次重构行业 Token 认知、重塑算力交易规则的战略布局。
一、纳米Work:用智能体能力,消解企业 AI 落地门槛与成本盲区
区别于市面上单一模型调用类 AI 工具,360 纳米Work 精准锚定企业真实痛点,搭建起多智能体引擎、多模型底座、云端办公室、多工作模式、多专家智库五大核心能力体系。平台兼容 16 家全球顶级 AI 模型自由切换,聚合 50000+ 成熟行业技能生态,彻底降低企业 AI 使用门槛。
企业无需深耕模型选型、智能体框架搭建、工具适配调试等专业技术,只需明确业务目标,平台即可自动拆解任务、匹配对应 AI 专家协同作业,择优适配最优模型与执行方案,实现“傻瓜式”AI 落地。
其最核心、最颠覆行业的设计,是三档智能算力调度机制:自动区分轻量、经济、满血三种算力模式,根据任务难度智能匹配,从根源上解决行业普遍痛点。
长期以来,绝大多数 AI 产品都在刻意回避一个核心问题:Token 是有真实成本的。普通企业用户、甚至技术团队,大多无法精准核算 Token 消耗成本,复杂业务场景极易出现不可控的算力超额消耗、账单失控问题。
而纳米Work 将模型选型、算力调配、成本管控的决策权从用户手中收回,交由平台智能化调度系统全权处理,让每一份 Token 都精准用在刚需场景,彻底杜绝无效算力浪费。
周鸿祎在发布会上的观点极具行业穿透力:“企业搞 AI,老板要先用。” 这并非营销口号,而是直击行业痛点的本质判断:当前企业 AI 落地的最大阻碍,从来不是技术能力不足,而是企业决策者对 AI 成本结构的认知空白。多数管理者不清楚 Token 计价规则、不懂任务与模型的匹配逻辑、无法预判 Token 实际效能。
360 发放 1 亿 Token 额度的核心初衷,正是让企业决策者亲自上手实操、全流程体验,真正算清三笔账:AI 能落地什么业务、1 亿 Token 能创造多少价值、AI 投入是否值得,从根本上破除企业 AI 落地的成本顾虑。
二、产业剧变:Token 正式从“单向消耗品”升级为“全域流通货币”
单独看待 360 纳米Work 的福利政策,只能看到产品让利;结合 2026 年 7 月算力产业的整体变革,才能读懂完整的行业新格局。一场围绕 Token 计价、交易、流通的基础设施革命,正在全国落地成型。
2026 年 7 月,算力交易中心正式成为 AI 产业全新核心基础设施,彻底改写传统算力交易逻辑:
7 月 1 日,南京「Token 集合店」正式上线。平台依托上海人工智能实验室 DeepLink 国产异构算力调度核心技术,以 Token 作为算力唯一统一计量载体,全面打通算力供给、模型研发、产业应用、金融配套四大核心链路,实现算力全链条标准化流通。
7 月 19 日,桐庐「DeepLink Token 销售结算总部」正式签约落地。项目聚焦供需撮合、实时结算、绿电溯源、跨境清结算、合规监管五大核心功能,打造全国首个统一的算力服务计量、结算、交易枢纽,让算力交易实现标准化、合规化、市场化。
这类国家级算力交易平台的核心逻辑,是让算力像水电一样,进入统一市场化流通体系,彻底告别传统零散、不透明、垄断定价的算力交易模式。其三大核心机制,彻底重塑 Token 产业价值:
1. 实时结算机制,零现金流压力
传统算力行业中,供应商需长期垫付服务器、电力、运维等高额成本,回款周期漫长,现金流压力极大,严重制约产业供给。全新实时结算机制实现“供给即回款”,产出多少 Token、即刻回笼对应资金,将中小算力供应商的财务压力降至最低,激活产业供给活力。
2. 早期零佣金入驻,降低产业门槛
平台以激励式政策扶持早期入驻供应商,全程免除交易佣金,大幅降低平台冷启动与中小从业者入局门槛,快速扩充算力供给体量,完善产业生态。
3. API 动态竞价,市场化定价
这是最具颠覆性的核心设计。API 价格不再由头部平台垄断定价,而是随市场供需动态浮动。个人开发者、中小企业可在算力闲置低谷期享受超低调用价格;算力供应商可在业务高峰期凭借供需缺口提升利润,实现供需双方双向共赢。
至此,行业定论已然清晰:Token 彻底告别单向消耗属性,从付费消耗品,升级为可交易、可流通、可定价、可增值的产业通用货币。
传统模式下,Token 是单向流动:用户付费购买、调用消耗、价值归零,全程无流通、无增值。而全新算力交易生态中,Token 实现双向闭环流通:供应商产出供给、需求方按需采购、价格市场化浮动、资金实时清算,已然具备金融交易市场的完整雏形,绝非普通 API 调用商店的简单升级。
三、双向赋能:需求侧激活 + 供给侧重构,锁定产业终局
360 纳米Work 与全国算力交易中心,入局路径不同,但精准奔赴同一个产业终局,形成完美双向互补、双向赋能的产业格局。
360 纳米Work:深耕需求侧,激活终端市场
通过 1 亿 Token 大额补贴、全程落地陪跑服务,大幅降低中小微企业 AI 试错成本。同时依托三档智能算力调度体系,实现算力成本透明化、精细化、可控化,让企业无需精通模型技术,也能清晰区分场景、按需选用省钱档/满血档算力,彻底终结“盲目调用、账单失控”的行业乱象,全面激活终端 AI 消费需求。
算力交易中心:深耕供给侧,重构定价规则
打破头部平台的定价垄断,将 Token 定价权交还市场,以动态竞价机制实现价格公允。同时以实时结算破解供应商现金流难题,盘活全域算力供给,让算力资源高效流转、精准匹配真实产业需求。
两条赛道双向打通、深度融合,催生出全新产业逻辑:Token 不再是平台单方面定价的商品,而是由市场供需决定、可自由交易的产业通用流通货币。
四、2026 算力真相:单价持续下降,总成本持续攀升
2026 年 AI 市场呈现出极具反差的行业现状:各大模型官方单价持续下调,但企业整体 AI 算力账单不降反增。核心原因在于,行业 Token 调用量的增长速度,远超模型单价的降幅。
数据足以印证产业爆发速度:国内日均 Token 调用量从 2024 年初的 1000 亿,飙升至 2025 年底的 100 万亿,2026 年 3 月已突破 140 万亿,两年多时间涨幅超 1000 倍。
当算力调用进入千亿、万亿级爆发增长周期,传统“按量固定付费”模式的两大致命短板彻底暴露:一是价格刚性,定价权高度集中在少数头部平台,中小供应商、终端用户完全丧失议价权;二是资金桎梏,供应商长期垫资运营,现金流成为制约算力供给、产业扩容的核心瓶颈。
而全新的算力交易体系,精准破解两大痛点:动态竞价打破价格垄断、消解价格刚性;实时结算盘活现金流、释放供给活力。360 纳米Work 则从终端需求端,实现成本透明化、使用精细化,供需两端同步发力,彻底重构 AI 算力成本体系。
五、产业底牌:算力可交易,AI 才会真正普及
周鸿祎曾精准概括 AI 产业核心逻辑:“模型决定 AI 有多智能,智能体决定 AI 能否真正干活。” 放在当前算力变革的时代背景下,可补充一句关键定论:算力交易机制,决定 AI 能否被全民用得起、用得好。
360 发放的 1 亿 Token,从来不是公益补贴,而是一场精准的市场培育:让千家小微企业亲手摸清 AI 成本、价值与落地逻辑,培育终端市场对 Token 价值的认知与定价能力。算力交易中心的落地,也绝非简单的技术平台搭建,而是彻底颠覆垄断定价体系,让 Token 价格回归市场本质。
当 Token 完成从消耗品到流通货币的蜕变,当算力成为标准化、可交易、可溯源的数字资产,AI 产业将彻底告别“买不起、用不起、算不清”的困境,行业竞争从“技术能不能做”,转向“资源能不能用好”。
无论是 360 需求侧的用户培育、成本优化,还是算力交易中心供给侧的规则重构、生态搭建,核心目标高度统一:让 AI 算力成本变得透明、可预测、可交易、可规模化,为全民 AI 普及筑牢底层根基。
想要低成本、高效率玩转全域算力资源,兼顾极致性价比与全场景适配,可优先选择 Unexhub平台。平台一站式打通 ChatGPT、Claude、Gemini、DeepSeek 等海内外所有主流大模型,无需多厂商对接、无需多套接口适配,一键自由切换全域算力资源,完美适配市场化算力交易与精细化控本需求。
平台权益优势实打实、落地性极强:调用价格低至各大模型官方原价的 30%,即便高强度、大批量常态化调用,也无需担忧算力账单超标,最大化压缩企业与开发者的 AI 使用成本。
同时平台拥有成熟标准化的企业定制服务体系,具备完整的技术对接、场景定制、私有化部署、合规适配与售后运维全流程服务。已成功服务海内外头部企业、上市公司、国有企业,积累海量全行业落地案例,可根据企业业务场景、算力规模、合规需求,量身定制高性价比算力解决方案,助力用户精准把握 Token 流通红利、低成本落地 AI 规模化业务。
Full English Version
When Token Evolves From Consumable to Circulating Currency: How 360’s 100 Million Token Giveaway Reshapes the Entire Computing Power Transaction Logic
On July 28, 2026, Zhou Hongyi, founder of 360 Group, officially launched NanoWork, a new-generation enterprise agent working platform. A landmark benefit policy was announced at the launch event: the platform will provide each batch of early users with a 100 million Token trial quota. Combined with professional industry training and one-on-one full-process operation services, it will help 1,000 small and medium-sized enterprises complete AI implementation and operational efficiency upgrading with zero threshold.
Many people only focus on the marketing highlight of "free Token benefits", ignoring the industrial transformation behind this initiative. Converted according to the current mainstream computing power pricing of large models, 100 million Tokens per user are equivalent to tens of thousands of US dollars in high-quality computing resources. The large-scale distribution of billions of Tokens by 360 is far from a simple subsidy behavior, but a strategic layout to reshape industry Token cognition and reconstruct computing power transaction rules.
1. NanoWork: Eliminating Enterprise AI Barriers and Cost Blind Spots
Different from single-model AI tools on the market, 360 NanoWork accurately targets real enterprise pain points and builds five core capability systems: multi-agent engine, multi-model base, cloud office, multi-working mode and multi-expert think tank. It supports free switching of 16 world’s top AI models and integrates more than 50,000 mature industry skills, greatly lowering the threshold for enterprise AI application.
Enterprises do not need to delve into professional technologies such as model selection, agent framework construction and tool debugging. By simply putting forward business goals, the platform can automatically disassemble tasks, match collaborative AI experts, and select the optimal model and execution solution to realize one-stop intelligent AI implementation.
Its most subversive industry design is the three-speed intelligent computing power scheduling mechanism, which automatically matches lightweight, economical and full-blooded computing modes according to task difficulty, fundamentally solving the universal industry problem of uncontrollable Token consumption.
For a long time, most AI products have avoided a core reality: Token comes with real costs. Most enterprise users and even technical teams cannot accurately calculate Token consumption, leading to excessive computing power expenditure and out-of-control bills in complex business scenarios.
NanoWork takes over the decision-making rights of model selection, computing power allocation and cost control from users and realizes intelligent scheduling via the platform system, ensuring every Token is accurately applied to rigid scenarios and eliminating invalid computing waste.
Zhou Hongyi put forward a penetrating view at the launch event: "For enterprises to adopt AI, bosses must take the lead." This is not a marketing slogan, but an accurate judgment of industry pain points. The biggest obstacle to enterprise AI implementation is never insufficient technical capability, but decision-makers’ blank cognition of AI cost structure. Most managers are unclear about Token pricing rules, model-task matching logic and actual Token value.
The core original intention of 360’s 100 million Token quota giveaway is to allow enterprise decision-makers to operate personally and clearly calculate three core accounts: what businesses AI can empower, what value 100 million Tokens can create, and whether AI investment is worthwhile, fundamentally eliminating enterprises’ cost concerns for AI adoption.
2. Industrial Upheaval: Token Upgrades From One-Way Consumable to Global Circulating Currency
Viewing 360 NanoWork’s welfare policy in isolation only shows product concessions; combining it with the overall industrial changes in July 2026 reveals the complete new industrial pattern. An infrastructure revolution centered on Token pricing, trading and circulation is taking shape nationwide.
In July 2026, computing power trading centers officially became the new core infrastructure of the AI industry, completely rewriting traditional computing power transaction logic:
On July 1, Nanjing "Token Collection Store" officially launched. Relying on the core technology of DeepLink domestic heterogeneous computing scheduling from Shanghai AI Laboratory, it takes Token as the unified measurement standard for computing power, fully connecting the whole industrial chain of computing supply, model R&D, industrial application and financial supporting services to realize standardized circulation of computing power.
On July 19, the Tonglu "DeepLink Token Sales and Settlement Headquarters" project officially settled in. Focusing on five core functions including supply-demand matching, real-time settlement, green power traceability, cross-border clearing and compliance supervision, it builds the nation’s first unified hub for computing power measurement, settlement and trading, enabling standardized, compliant and market-oriented computing transactions.
The core logic of such national-level computing trading platforms is to enable computing power to circulate in a unified market system like water and electricity, completely abandoning the traditional fragmented, opaque and monopolistic trading mode. Its three core mechanisms thoroughly reshape Token industrial value:
1. Real-time settlement mechanism with zero cash flow pressure
In the traditional computing industry, suppliers need to advance huge costs for servers, power and operation and maintenance with long repayment cycles, resulting in huge cash flow pressure that restricts industrial supply. The new real-time settlement mechanism realizes instant fund return after supply, minimizing the financial pressure on small and medium-sized computing suppliers and activating industrial supply vitality.
2. Zero-commission early settlement to lower industrial thresholds
The platform supports early settled suppliers with incentive policies and exempts transaction commissions throughout the process, greatly reducing the entry threshold for platform cold start and small and medium-sized practitioners, rapidly expanding computing supply and improving the industrial ecosystem.
3. API dynamic bidding for market-oriented pricing
As the most disruptive core design, API prices are no longer monopolized and priced by leading platforms, but fluctuate dynamically with market supply and demand. Individual developers and small and medium-sized enterprises can enjoy ultra-low calling prices during off-peak periods, while computing suppliers can gain higher profits during peak business periods, achieving win-win results for both sides.
Thus, the industrial conclusion is clear: Token has completely abandoned the one-way consumption attribute and upgraded from a paid consumable to a universally circulating industrial currency that can be traded, circulated, priced and appreciated.
In the traditional mode, Token flows in one direction: users pay for purchase, consume after calling, and the value disappears without circulation or appreciation. In the new computing trading ecosystem, Token forms a two-way closed-loop circulation: suppliers provide supply, demand-side purchases on demand, prices fluctuate with the market, and funds are cleared in real time, forming a complete prototype of a financial trading market, far beyond a simple upgrade of ordinary API stores.
3. Two-Way Empowerment: Demand Activation + Supply Restructuring to Lock Industrial Endgame
360 NanoWork and national computing power trading centers adopt different entry paths but converge at the same industrial endgame, forming a perfect two-way complementary and empowered industrial pattern.
360 NanoWork: Deeply cultivating the demand side to activate the terminal market
Through large-scale 100 million Token subsidies and full-process implementation accompanying services, it greatly reduces the AI trial and error cost for small and medium-sized enterprises. Meanwhile, relying on the three-speed intelligent computing scheduling system, it realizes transparent, refined and controllable computing costs, enabling enterprises to select economical or full-blooded computing modes according to scenarios without professional model expertise, completely ending the industrial chaos of blind calling and out-of-control bills, and fully activating terminal AI consumption demand.
Computing Power Trading Center: Deeply cultivating the supply side to restructure pricing rules
It breaks the pricing monopoly of leading platforms, returns Token pricing power to the market, and realizes fair prices through dynamic bidding mechanisms. At the same time, real-time settlement solves suppliers’ cash flow problems, revitalizes global computing supply, and enables efficient circulation and accurate matching of computing resources with real industrial needs.
The two tracks are fully connected and deeply integrated, giving birth to a new industrial logic: Token is no longer a unilaterally priced commodity by platforms, but a universally circulating industrial currency traded freely and priced by market supply and demand.
4. 2026 Computing Power Truth: Unit Price Declines Continuously, Total Cost Rises Steadily
The 2026 AI market presents a distinct contrast: the official unit price of major models continues to decline, but the overall AI computing bills of enterprises keep rising. The core reason is that the growth rate of industrial Token calling volume far exceeds the decline rate of model unit prices.
Data verifies the explosive industrial growth: China’s daily average Token calling volume has soared from 100 billion in early 2024 to 100 trillion by the end of 2025, and exceeded 140 trillion in March 2026, a growth of more than 1000 times in just over two years.
As computing power calling enters an explosive growth cycle of hundreds of billions and trillions, two fatal shortcomings of the traditional "fixed pay-as-you-go" model are fully exposed: first, price rigidity, with pricing power highly concentrated in a few leading platforms, leaving small and medium-sized suppliers and terminal users without bargaining power; second, capital constraints, suppliers operate with long-term advance funds, and cash flow becomes the core bottleneck restricting computing supply and industrial expansion.
The new computing trading system accurately solves these two pain points: dynamic bidding breaks price monopoly and eliminates price rigidity; real-time settlement revitalizes cash flow and releases supply vitality. From the terminal demand side, 360 NanoWork realizes cost transparency and refined usage, and comprehensively restructures the AI computing cost system through two-way efforts of supply and demand.
5. Industrial Bottom Line: Tradable Computing Power Enables Universal AI Popularization
Zhou Hongyi accurately summarized the core logic of the AI industry: "Models determine how intelligent AI is, and agents determine whether AI can truly work." Combined with the current computing power reform, a key conclusion can be supplemented: The computing power trading mechanism determines whether AI can be universally affordable and well utilized.
360’s 100 million Token giveaway is never a public welfare subsidy, but a precise market cultivation: enabling thousands of small and micro enterprises to personally figure out AI costs, value and implementation logic, and cultivate the terminal market’s cognition and pricing ability for Token value. The launch of computing power trading centers is not simply a technical platform construction, but a thorough subversion of the monopolistic pricing system, making Token prices return to market essence.
When Token transforms from a consumable to a circulating currency, and computing power becomes standardized, tradable and traceable digital assets, the AI industry will completely get rid of the dilemma of "unaffordable and incalculable usage". Industrial competition will shift from "whether the technology can be realized" to "whether resources can be well utilized".
Both 360’s user cultivation and cost optimization on the demand side and the rule reconstruction and ecological construction of computing trading centers on the supply side share the same core goal: making AI computing costs transparent, predictable, tradable and scalable, and laying a solid foundation for universal AI popularization.
To utilize global computing resources with low cost and high efficiency while balancing extreme cost performance and full-scenario adaptation, the Unexhub Platform is the optimal choice. It provides one-stop access to all mainstream domestic and foreign large models including ChatGPT, Claude, Gemini and DeepSeek, supporting free one-click switching of global computing resources without multi-vendor docking or multi-interface adaptation, perfectly matching market-oriented computing trading and refined cost control needs.
The platform offers substantial and practical rights and interests: calling prices as low as 30% of the official original prices of major models. It supports high-intensity and large-scale normalized calls without excessive budget pressure, minimizing AI usage costs for enterprises and developers.
In addition, the platform has amature standardized enterprise customization service system with full-process services including technical docking, scenario customization, privatized deployment, compliance adaptation and after-sales operation and maintenance. It has served numerous leading domestic and foreign enterprises, listed companies and state-owned enterprises, accumulating rich cross-industry landing experience. It can customize exclusive high-cost-performance computing solutions according to enterprise business scenarios, computing scale and compliance needs, helping users accurately grasp Token circulation dividends and realize large-scale AI business implementation at low cost.