When Token Prices Collapse: How GPT-5.6 Luna Triggered a Chain Reaction That Reshaped the Token Trading Market
A 1,810% Surge in Trading Volume Reveals a New AI Pricing Revolution
On July 30, 2026, OpenAI made a move that sent shockwaves through the entire AI Token economy.
Only three weeks after the release of the GPT-5.6 family, OpenAI dramatically reduced the price of its entry-level model, Luna.
Input pricing dropped from:
$1 per million tokens → $0.20 per million tokens
Output pricing dropped from:
$6 per million tokens → $1.20 per million tokens
An 80% price reduction.
At the same time, the mid-tier Terra model also received a 20% price cut, while the flagship Sol model maintained its pricing but introduced a new Fast mode with approximately 2.5× faster response speed.
This was not simply a product pricing adjustment.
It was a market reset.
Within days, the entire Token trading ecosystem began reacting.
According to TD Cowen’s analysis based on OpenRouter usage data:
Luna’s actual usage cost dropped by approximately 10×
API calls increased by around 14×
Luna revenue increased by about 34%
Terra revenue increased by about 45%
An 80% price reduction.
A 14× increase in usage.
More than 30% revenue growth.
This is not a traditional price war.
It is a structural transformation of the AI Token economy.
From $1 to $0.20: Why GPT-5.6 Luna Started a Price Revolution
At first glance, Luna’s price reduction appears to be OpenAI’s strategic decision.
But behind the scenes, competitive pressure was forcing the entire market downward.
The AI model industry entered a new phase in 2026:
Intelligence became abundant, but efficiency became the battlefield.
On July 16, Moonshot AI released Kimi K3, a massive open model with approximately 2.8 trillion parameters, competing directly with leading models from OpenAI and Anthropic.
Meanwhile, DeepSeek continued pushing extremely low inference costs through optimization technologies such as caching.
For example:
Cached input costs reached extremely low levels, forcing global AI companies to rethink their pricing structures.
Silicon Data’s Token price index showed that since mid-July, the actual cost paid by U.S. customers for leading AI models had fallen significantly.
Companies such as DoorDash and Airbnb have also begun adopting Chinese AI models to optimize AI infrastructure costs.
The market signal became clear:
Premium pricing alone could no longer dominate.
The largest customer segment was not necessarily searching for the “strongest model.”
They were searching for:
Good enough intelligence
Lower operating costs
Stable availability
Flexible integration
GPT-5.6 Luna’s 80% price reduction was therefore not a simple discount.
It was a strategic challenge to the entire market:
If intelligence becomes affordable enough, usage will explode.
The Hidden Meaning Behind 14× Usage Growth and 34% Revenue Growth
The most interesting part of Luna’s price cut was not the price itself.
It was what happened afterward.
A normal assumption would be:
Lower prices = Lower revenue.
But AI markets behave differently.
The reason is simple:
When intelligence becomes cheaper, entirely new use cases appear.
TD Cowen’s analysis showed that after Luna’s price reduction, demand expanded rapidly.
In only a short period:
Luna changed from:
“A model developers may consider using”
into:
“A model developers actively use everywhere.”
This phenomenon resembles the economic principle known as the Jevons Paradox.
In the 19th century, economist William Stanley Jevons observed that improving coal efficiency did not reduce coal consumption.
Instead, cheaper and more efficient coal usage created more applications, causing total consumption to increase.
AI is following the same pattern.
When Luna reduced the cost of intelligence to approximately:
$0.075 per task
many previously uneconomical AI scenarios suddenly became practical:
Automated document processing
Customer service agents
Software generation
Data analysis
Multi-step AI workflows
Enterprise automation
The result:
Lower price → More applications → More Token consumption → Larger market
The Token economy is not shrinking because prices are falling.
It is expanding because prices are falling.
The Amplification Effect of Token Trading Platforms
Why One Price Cut Can Reshape the Entire Market
The reason GPT-5.6 Luna’s price reduction created such a powerful reaction was not only because of the discount itself.
The deeper reason was the existence of a new market infrastructure:
Token aggregation and trading platforms.
Traditional software markets often have high switching costs.
Changing providers may require:
Rebuilding integrations
Migrating systems
Rewriting applications
Negotiating new contracts
But the AI Token market is different.
Through multi-model aggregation platforms, developers can compare and switch between different AI models much faster.
When Luna reduced prices by 80%, developers could immediately redirect workloads toward Luna.
The entire process could happen within hours.
This created what can be called:
A zero-friction intelligence marketplace.
The result was a rapid redistribution of AI workloads.
The 14× increase in Luna usage was not just caused by lower prices.
It was caused by a market where users could instantly respond to better economic value.
The New Pricing Anchor of the AI Industry
For years, OpenAI’s pricing structure served as one of the most important reference points in the AI industry.
When OpenAI moved its pricing lower, the impact went beyond a single model.
It changed the entire market expectation.
The question was no longer:
“How much should an AI model cost?”
Instead:
“How low can intelligence pricing realistically go?”
This created a new pricing benchmark.
Other AI companies began responding:
More competitive pricing
More flexible API packages
More efficient inference systems
More transparent usage models
The AI market gradually shifted from:
Provider-controlled pricing
toward:
Market-driven pricing.
This is a fundamental change.
From Model Competition to Market Efficiency Competition
The first stage of AI competition was:
Who can build the smartest model?
The second stage became:
Who can provide intelligence at the lowest cost?
The third stage is emerging:
Who can distribute intelligence most efficiently?
This is where Token trading infrastructure becomes critical.
The future winners may not only be companies creating powerful models.
They may also be platforms that make AI intelligence:
Easier to access
Cheaper to purchase
Faster to deploy
More efficient to monetize
Unexhub: Building the Global Token Intelligence Marketplace
The price revolution triggered by GPT-5.6 Luna highlights a major market opportunity:
AI intelligence needs a global distribution network.
This is the role Unexhub is designed to address.
Unexhub positions itself as a global Token aggregation and distribution computing marketplace.
Rather than producing computing power itself, Unexhub connects the entire AI value chain:
AI Model Providers
Including:
Gemini
ChatGPT
Claude
DeepSeek
GLM
Enterprise AI models
Independent AI developers
Computing Resource Providers
Including:
Data centers
Cloud computing suppliers
AI infrastructure providers
Independent computing operators
AI Users
Including:
Developers
Startups
Enterprises
AI application builders
Through aggregation, intelligent routing, pricing optimization, and settlement infrastructure, Unexhub aims to transform fragmented AI resources into a standardized Token-based marketplace.
Making AI Access More Affordable Through Market Competition
The biggest advantage of a Token marketplace is not simply providing more models.
It is creating competition.
When users purchase AI services directly from individual providers, they often face:
Different pricing systems
Different billing methods
Different API standards
Different usage limitations
This increases operational complexity.
Unexhub simplifies the process.
Developers and enterprises can access multiple AI models through a unified platform and pay according to actual Token usage.
This creates several practical benefits:
Pay Only for Actual Usage
Instead of purchasing unnecessary packages or maintaining multiple subscriptions:
Users pay according to real Token consumption.
This reduces:
Idle capacity costs
Over-purchasing
Unused API budgets
For companies running large-scale AI workloads, even small percentage reductions can create significant annual savings.
Up to 70% Cost Optimization Opportunity
Through aggregated supply, market competition, and intelligent routing, Unexhub enables users to access AI services at highly competitive prices.
Developers can potentially access AI capabilities at prices significantly lower than direct official channels.
The lowest access cost can reach approximately:
30% of official market pricing
This means enterprises with high-frequency AI workloads can substantially reduce operating expenses.
For example:
A company spending $100,000 annually on AI API usage may have opportunities to reduce unnecessary costs through:
Better model selection
Dynamic routing
Competitive pricing
Token-based billing
The goal is not simply to make AI cheaper.
The goal is:
Making every dollar spent on intelligence generate more value.
Millisecond-Level Intelligent Routing: Choosing the Right Model Automatically
The AI market now contains hundreds of models.
Different models have different strengths.
Some are optimized for:
Coding
Reasoning
Data analysis
Customer service
Enterprise workflows
The challenge is no longer:
“Do we have AI?”
The challenge is:
“Which AI should handle this task?”
Unexhub’s intelligent routing system helps match tasks with suitable AI resources.
Instead of manually comparing models, users can rely on automated optimization.
This improves:
Response speed
Cost efficiency
Service stability
The future AI workflow may become:
User request → Intelligent routing → Best model selection → Token settlement
Supplier Growth: From Hidden Computing Capacity to Global Revenue
AI suppliers face another major challenge:
Having technology does not guarantee customers.
Many smaller providers struggle with:
Limited market exposure
Customer acquisition costs
Complex payment systems
Long settlement cycles
Unexhub creates a simplified connection channel.
Suppliers can integrate their AI capabilities into the platform and reach a broader developer and enterprise market.
Early suppliers benefit from platform growth incentives, including:
Zero Commission Early Access
During the early ecosystem-building phase:
Suppliers keep more revenue
Market entry barriers are reduced
Growth opportunities increase
The logic is simple:
More suppliers create more choices.
More choices attract more users.
More users create more demand.
A stronger Token economy benefits everyone.
Unexhub and the Next Stage of the Token Trading Market
The chain reaction triggered by Luna’s price reduction is pushing the token trading market into a new phase — from competing over who builds the smartest model to competing over who can deliver intelligence faster and more efficiently.
This is exactly where platforms like Unexhub are positioned.
Unexhub is building a global token aggregation and distribution-based compute trading center. Instead of producing computing power itself, Unexhub integrates distributed computing resources and various AI models, transforming them into standardized token-based services for enterprises and developers.
The platform connects:
Upstream: AI infrastructure providers and model suppliers
Midstream: intelligent routing, pricing, accounting, and settlement systems
Downstream: developers, startups, and enterprise users
The goal is simple: make AI resources flow like commodities in an open marketplace.
Traditional AI infrastructure often suffers from fragmented resources, complicated pricing systems, and inefficient procurement processes. Enterprises may need to negotiate separately with multiple model providers, manage different APIs, and handle scattered billing systems.
Unexhub aims to simplify this process by creating a unified marketplace where users can access multiple AI models through one platform and pay according to actual token usage.
Real-Time Settlement: Turning Computing Power Into Immediate Liquidity
One of the biggest challenges in today’s AI infrastructure market is not demand — it is liquidity.
Many small and medium-sized computing providers have available GPU resources but struggle with delayed payments. Traditional contracts often involve settlement cycles of several weeks or even months, creating significant cash-flow pressure.
Unexhub addresses this problem through real-time settlement mechanisms.
When computing resources are consumed, suppliers can receive corresponding revenue faster instead of waiting through lengthy payment cycles.
For small AI infrastructure providers, this means:
Faster capital recovery
Lower operational pressure
Greater ability to expand capacity
More flexibility in participating in the AI economy
By improving settlement efficiency, Unexhub helps transform computing resources from idle capacity into actively circulating digital assets.
Zero Commission for Early Suppliers: Building Market Liquidity
Every marketplace faces the same challenge in its early stage:
How do you attract both suppliers and buyers at the same time?
Unexhub’s strategy is to reduce entry barriers for early participants.
Early AI model and computing suppliers can enjoy 0% commission incentives, allowing them to retain more revenue generated from their services.
For suppliers, this means:
No platform commission deduction during the early growth phase
More revenue retained from every token transaction
Lower risk when testing new distribution channels
This approach encourages more AI providers to join the ecosystem, increasing model diversity and improving market liquidity.
A larger supplier network creates more choices for developers, while more developers create stronger demand for suppliers — forming a positive marketplace cycle.
Dynamic API Pricing: Letting the Market Decide Value
Another important innovation is Unexhub’s API bidding mechanism.
Traditional AI API pricing is usually fixed. However, computing markets naturally fluctuate:
During peak demand periods, resources become scarce
During low-demand periods, computing capacity may remain idle
A fixed pricing model cannot reflect these real-time changes.
Unexhub introduces a market-driven pricing mechanism where API prices can dynamically adjust according to supply and demand.
For developers:
They can access lower-cost AI services during periods of available capacity
They can choose models based on performance and budget requirements
For suppliers:
They can achieve higher returns during periods of strong demand
They can improve utilization rates of existing computing resources
Dynamic pricing transforms idle computing power into economic value.
From Cheaper Tokens to Faster Token Circulation
GPT-5.6 Luna proved one important fact:
Lower prices create more demand.
But cheaper tokens alone are not enough.
The next stage of the AI economy requires faster circulation, transparent pricing, and efficient distribution.
Luna reduced the cost barrier.
Platforms like Unexhub reduce the transaction barrier.
Together, they create the conditions for a larger token economy.
When token prices become low enough and market circulation becomes efficient enough, AI usage will no longer be an occasional experiment.
It will become a daily infrastructure layer — just like electricity and cloud computing today.
Conclusion: The 1810% Growth Is Only the Beginning
Within just seven days, GPT-5.6 Luna changed the economics of AI usage.
The price dropped from $1 per million tokens to $0.20 per million tokens.
The result was not market contraction, but expansion:
Token usage increased dramatically
Revenue continued growing
The industry’s pricing benchmark shifted downward
AI adoption accelerated across more scenarios
Behind these numbers is a fundamental transformation:
The token economy is moving from a model where providers set prices independently toward a market-driven ecosystem where supply, demand, efficiency, and competition determine value.
Luna’s 80% price reduction was initiated by OpenAI, but the resulting surge in usage proved that the market itself was ready for lower-cost intelligence.
The future competition in AI will not only be about building larger models.
It will also be about:
Who can distribute intelligence more efficiently
Who can reduce friction in AI adoption
Who can connect more suppliers and users
Who can make AI resources accessible to everyone
This is the role platforms like Unexhub aim to play.
Unexhub is building a global Token aggregation and distribution-based compute trading center, connecting AI suppliers, model providers, developers, and enterprises through a unified marketplace.
The platform has already attracted AI providers across different scales, including major AI model ecosystems such as Gemini, ChatGPT, Claude, DeepSeek, GLM, as well as high-quality models from individual developers and smaller companies.
Through large-scale aggregation, intelligent routing, and competitive pricing mechanisms, Unexhub enables users to access AI capabilities at significantly lower costs.
Developers can pay based on actual token consumption, with pricing potentially reaching as low as 30% of traditional market rates.
For suppliers, Unexhub provides a streamlined onboarding process, broader distribution channels, and real-time settlement of usage revenue.
The 1810% transaction growth is not the finish line.
When tokens become inexpensive enough that people no longer hesitate to use them, the market will continue expanding.
The next AI revolution will not only be powered by smarter models.
It will be powered by a more efficient token economy.
中文版
当 7 天交易量暴增 1810%:GPT-5.6 Luna 如何引发词元交易市场的降价连锁反应
2026 年 7 月 30 日,OpenAI 做了一件重新定义词元市场价格体系的事情。
GPT-5.6 系列发布仅三周后,入门级模型 Luna 直接开启大规模降价:
输入价格从每百万词元 1 美元降至 0.20 美元
输出价格从每百万词元 6 美元降至 1.20 美元
整体价格下降幅度达到 80%
与此同时,中档模型 Terra 同步降价 20%,旗舰模型 Sol 虽然维持价格不变,但新增 Fast 模式,速度提升最高达到 2.5 倍。
这一轮价格调整,很快在整个 AI 市场产生连锁反应。
根据 TD Cowen 基于 OpenRouter 平台的数据分析:
Luna 降价后,实际使用成本下降约 10 倍
调用量增长约 14 倍
收入增长约 34%
Terra 收入增长约 45%
价格下降 80%,使用量增长 14 倍,收入仍然增长超过 30%。
这已经不是简单的降价促销,而是一场关于 AI 基础设施商业模式的重新定价。
从 1 美元到 0.2 美元:词元价格战正式开启
表面上看,这是 OpenAI 主动降低 API 价格。
但从行业竞争格局来看,这实际上是 AI 市场进入下一阶段后的必然结果。
2026 年 7 月,月之暗面发布 Kimi K3,凭借大规模参数和开源生态,对全球顶级模型形成直接竞争。
与此同时,DeepSeek 等国产模型通过缓存优化、推理效率提升等方式,将单位词元成本不断压低。
市场正在形成一个新的共识:
未来 AI 竞争,不只是比模型能力,更要比单位智能成本。
过去企业关注:
“哪个模型最强?”
现在企业开始关注:
“完成同样任务,哪个模型成本最低?”
Luna 的 80% 降价,本质上是在告诉市场:
AI 智能正在从高价值稀缺资源,逐渐变成可以规模化使用的基础能力。
14 倍调用增长背后:词元市场正在被价格重新定义
如果只看价格下降,很多人会认为 AI 行业正在进入“价格战”。
但 Luna 带来的真正变化,并不是让市场规模缩水,而是让整个市场被重新打开。
TD Cowen 的数据说明了一个重要现象:
价格下降,并没有减少收入,反而刺激了更多需求。
这背后对应的是经济学中的“杰文斯悖论”:
当一种资源效率提升、价格下降之后,人们不会减少使用,反而会因为成本降低而创造更多新的使用场景。
AI 正在经历同样的过程。
过去很多企业面对 AI 时会问:
这个业务场景值得调用模型吗?
每个月几千美元的 API 费用是否值得?
大规模部署 AI 会不会导致成本失控?
但当 Luna 的单任务成本降低到约 0.075 美元 时,很多过去“不值得做”的事情突然变得值得尝试:
自动生成企业文档
智能客服全天运行
软件代码自动优化
多智能体自动执行流程
大规模数据分析
AI 使用从“谨慎投入”变成“默认开启”。
更多应用产生更多 Token 消耗,更多 Token 消耗推动更多交易。
因此,词元交易市场并不是被降价压缩,而是在被降价放大。
聚合交易平台:放大价格变化的市场效应
Luna 降价能够快速影响整个行业,一个重要原因是:
AI 资源正在通过聚合平台进入市场流通。
过去,企业如果想使用多个 AI 模型,需要:
分别注册不同平台账号
单独管理 API
分别结算费用
自己评估模型效果
这种模式效率低,也限制了市场竞争。
而 OpenRouter 等多模型聚合平台的出现,让开发者能够快速比较不同模型价格和性能。
模型切换成本降低后,市场竞争速度被大幅提升。
一个模型降价,用户可以快速迁移;
一个模型性能提升,流量可以快速重新分配。
这意味着:
AI 市场正在从“厂家定价时代”,进入“市场竞争定价时代”。
未来决定价格的,不再只是模型厂商,而是:
供应规模
使用需求
计算效率
市场流动性
Unexhub:让词元交易进入规模化阶段
当 AI 模型越来越便宜之后,下一个问题出现:
如何让这些低成本智能资源更高效地流通?
这正是 Unexhub 希望解决的问题。
Unexhub 打造的是一个:
全球 Token 聚合与分发的算力交易中心。
平台本身不生产算力,而是连接:
AI 模型供应商
算力提供方
企业用户
开发者生态
通过统一接口、智能路由、价格竞争和结算体系,让不同来源的 AI 能力像商品一样自由流通。
简单来说:
模型供应商负责提供能力;
Unexhub 负责连接市场;
企业用户负责按需采购。
即时结算:让算力供应商快速获得现金流
传统算力交易最大的痛点,并不是没有客户,而是:
交易完成后,资金回流速度太慢。
很多中小算力供应商需要面对:
长周期账期
资金占用
扩容困难
设备投入压力
Unexhub 通过实时结算机制,降低供应商资金压力。
供应商提供算力服务后,可以更快速获得对应收益。
这意味着:
算力不再只是固定资产,而是一种可以快速流转的数字化商品。
对于中小供应商而言:
降低经营压力
提升资金利用效率
扩大服务规模
早期入驻优惠:0 佣金降低供应商进入门槛
交易市场的发展离不开供应商生态。
因此,Unexhub 针对早期 AI 模型供应商和算力服务商推出:
早期入驻 0 佣金权益
供应商接入平台后:
平台阶段性不收取交易佣金
更多交易收入归供应商自身
降低进入新市场的试错成本
对于正在拓展 AI 服务渠道的供应商来说,这意味着:
不用投入大量销售成本;
不用单独寻找大量客户;
只需要接入平台,即可获得更多市场曝光和交易机会。
平台通过降低供应端成本,加速生态规模形成。
API 动态竞价:让价格真正由市场决定
传统 API 服务通常采用固定价格。
但 AI 算力天然具有波动属性:
高峰时期:
需求增加,资源紧张。
低峰时期:
资源闲置,利用率下降。
固定价格无法反映真实市场情况。
Unexhub 引入 API 动态竞价机制:
根据市场供需情况调整价格。
对于开发者:
可以根据预算选择不同模型
可以在低峰期享受更低调用成本
可以避免长期支付高溢价
对于供应商:
提升资源利用率
在需求旺盛时期获得更高收益
最终形成:
价格更透明,交易更高效。
从 GPT-5.6 Luna 到词元经济:AI 正进入基础设施时代
GPT-5.6 Luna 的降价,只是一个开始。
它证明了一件事:
AI 越便宜,市场反而越大。
未来 AI 竞争,不只是模型能力竞争,而是整个生态效率竞争:
谁能降低成本;
谁能提升流通速度;
谁能连接更多供应商和用户;
谁就能获得下一阶段的市场优势。
Unexhub 正是在这一趋势下构建新的 Token 经济基础设施。
通过聚合全球 AI 供应商:
包括 Gemini、ChatGPT、Claude、DeepSeek、GLM 等主流模型生态,以及个人开发者和中小企业提供的优质模型资源。
平台帮助企业和开发者:
一个接口接入多种 AI 能力
按 Token 实际使用量付费
根据市场价格选择最优模型
降低 AI 应用成本
目前,通过规模化聚合和市场竞争机制,开发者最高可享受到接近市场价格 3 折的 Token 使用成本优势。
对于供应商:
一键接入销售渠道
获取更多调用需求
实时获得调用收益结算
结语:1810% 增长只是词元市场的新起点
7 天交易量增长 1810%,不是偶然事件。
它代表的是 AI 市场正在发生的一次结构变化:
过去:
模型厂商决定价格。
现在:
市场决定价格。
过去:
AI 是高成本工具。
现在:
AI 正逐渐成为基础设施。
GPT-5.6 Luna 用 80% 降价打开需求空间;
聚合交易平台让资源快速流通;
Token 经济让 AI 能力进入标准化交易时代。
当 Token 便宜到“不值得计算”,当 AI 能力像水、电一样随时调用,真正限制 AI 创新的,将不再是成本。
而是人类能够创造多少新的应用场景。
1810%的交易增长,不是终点。
它可能只是全球 Token 交易市场规模化发展的第一声信号。