Galxe: How a Quest Platform Evolved into Web3's Growth Infrastructure

marsbitPubblicato 2026-05-25Pubblicato ultima volta 2026-05-25

Introduzione

Galxe, once perceived as a simple Web3 quest platform, has evolved into a core growth infrastructure within the Web3 ecosystem. It addresses a fundamental Web3 growth dilemma: the lack of a mature, systematic user acquisition and retention system akin to Web2's advertising and analytics platforms. While users complete quests (social tasks, on-chain interactions) for rewards, Galxe's true innovation lies in transforming these fragmented, one-off actions into lasting, verifiable identity credentials. This process of *behavioral assetization* creates a persistent record of a user's activities across projects and chains. For users, their wallet accumulates a valuable history that can unlock future access and rewards, fostering a "profile-building" mentality. For projects, Galxe provides a pre-screened user pool with rich behavioral data, enabling targeted outreach to users based on their specific on-chain history and community engagement. Galxe employs a gamefied growth path, guiding users from low-friction social tasks into deeper, valuable on-chain interactions through a structured progression of quests. This solves the incentive-behavior mismatch common in Web3, filtering users by their willingness to engage. Beyond quests, products like Passport (identity verification) and Starboard (community analytics) position Galxe as a comprehensive growth operating system. The platform's defensible advantage is its self-reinforcing data and network flywheel: more projects attract mor...

Author:137Labs

Many people's first encounter with Galxe leads them to understand it as a typical Web3 Quest platform: users complete tasks like following Twitter, joining Discord, or performing on-chain interactions, then claim NFTs, points, or airdrop qualifications. Superficially, this logic doesn't differ fundamentally from the many task platforms that have emerged in recent years. Even in product form, Galxe's pages appear very "light," resembling more of a standardized activity tool. However, when one truly observes the growth trajectory of Web3 over the past few years, a thought-provoking phenomenon emerges: whether it's Optimism, Arbitrum, Linea, or new ecosystems like Berachain and Movement Labs, almost all have used Galxe as a core growth platform at some point. In other words, Galxe is not a marginalized tool but has gradually become part of the growth infrastructure within the Web3 ecosystem.

This also means that what Galxe truly provides is not merely "complete tasks, claim rewards," but a more fundamental capability: it is gradually productizing, systematizing, and data-fying the originally highly fragmented, short-cycled, and non-reusable growth processes of Web3.

The Growth Dilemma of Web3

Looking back at the past decade of internet development, one finds that the most mature capability in the Web2 world is not product development but the growth system. Facebook Ads, Google Ads, recommendation algorithms, user profiles, membership systems—these elements together form a complete, industrialized system for traffic. Any internet company can acquire users, screen them, and continuously optimize conversion and retention through advertising platforms, data analytics, and recommendation algorithms.

But the Web3 world has long lacked this capability.

Most Web3 projects, despite having tokens, communities, and on-chain data, have always been missing a mature infrastructure for user growth. It's difficult for project teams to know who are real users and who are merely airdrop hunters; there's no unified identity system, nor cross-platform user profiles; a large portion of growth methods still rely on Twitter, Discord, airdrops, and community referrals. Consequently, the industry has gradually fallen into a typical dilemma: projects can quickly gain traffic through incentives but find it hard to truly retain long-term users.

The emergence of Galxe, in essence, is filling this missing layer of "growth infrastructure." Galxe was originally named Project Galaxy, founded in 2021. Its core vision wasn't simply to be an activity platform but to establish an open Credential Data Network, aiming to help developers and project teams identify user identities through on-chain and off-chain behaviors. In 2022, Project Galaxy officially rebranded as Galxe. This upgrade wasn't merely a visual change but signaled a shift in its positioning—from a single product to gradually evolving into a complete ecosystem built around identity, growth, and distribution.

The Formation of the Founding Team and Product Path

The two core founders of Galxe, Harry Zhang and Charles Wayn, are not typical Crypto protocol entrepreneurs in the traditional sense. They previously co-founded the live-streaming platform DLive, which itself was a product highly reliant on community, creator incentives, and user growth. Harry Zhang had also been involved in projects like Lino Network. Therefore, they possess a strong internet product mindset regarding "how communities grow" and "why users stay."

This is also why Galxe, from the very beginning, didn't resemble a pure on-chain protocol but more like an internet growth product. It has a very distinct gamification structure: leveling systems, ranks, identity, points, task sequences, and continuous incentives—all mechanisms derived from proven growth experiences in the Web2 world. In a sense, what Galxe is doing is reintroducing Web2's growth logic into Web3.

Compared to many Web3 projects that emphasize "protocols," "decentralization," or "technical architecture," Galxe focuses more on user behavior itself. It doesn't attempt to change users through complex mechanisms but drives the conversion of users from observers to participants, and then to long-term retention, through lower barriers to entry, more continuous task structures, and clearer feedback mechanisms. Consequently, Galxe's subsequent product evolution path has consistently revolved around the same core: how to enable user behavior to be continuously recorded, verified, and reused.

Mechanism Analysis: Assetizing User Behavior

When many analyze Galxe, they easily focus on the Quest itself, as it's the most direct product form users see: project teams publish tasks, users complete actions like follows, retweets, community joins, or on-chain interactions, then receive NFTs, points, whitelist spots, or airdrop qualifications. However, if one stops at this level, it leads to understanding Galxe as a mere "task outsourcing tool," overlooking its true growth logic.

The key to Galxe is not getting users to complete a single, isolated task, but transforming these originally dispersed, short-term, non-reusable user actions into long-term identity data that can be recorded, verified, screened, and reused. In other words, the Quest is merely the entry point for users into the system; what truly settles is the user's behavioral resume across different projects, different chains, and different scenarios.

In traditional Web3 growth, airdrops and tasks often create a problem: users come for the reward, leave after completing the action, and project teams ultimately gain short-term data, not long-term relationships. For instance, a user might join a Discord for an airdrop today and complete a transaction for a whitelist tomorrow. After these tasks end, the behaviors often cease to generate value, and project teams find it difficult to judge whether the user is a genuine contributor, a short-term farm-bot, or a potential core user.

Galxe's approach is to turn each behavior into cumulative records—Credentials, OATs, Passports, Scores—ensuring user actions are not one-time consumables but enter a long-term identity account system. After completing a task, a user doesn't just "claim a reward"; they also gain an on-chain or off-chain record that can be showcased, verified, and invoked by subsequent activities.

This mechanism alters the user's psychological accounting of participation. In the past, users doing tasks were essentially performing growth actions for project teams. Within Galxe's system, while users complete tasks, they are also continuously enriching their identity records. A wallet that has participated in activities for ecosystems like Optimism, Linea, and Arbitrum likely holds a different weight compared to a brand-new empty wallet when it comes to future qualification acquisition, event access, or project recognition. Consequently, users gradually develop a "wallet nurturing" mindset: the richer my wallet's history, the more complete my participation records, and the more identity credentials I have, the higher my probability of gaining future benefits.

More importantly, this assetization of behavior serves not only users but also project teams. For project teams, what Galxe provides is not simple traffic, but a pool of users with tags, history, and screening capabilities. Project teams can filter users more aligned with their target audience based on their past on-chain interactions, community behavior, task completion status, and identity credentials. For example, a DeFi project might focus more on wallets that have used cross-chain bridges, DEXs, or lending protocols; a new L1 chain might want to find users who have participated in testnets, completed developer tasks, or have high-activity records; an NFT project might value collection history, community activity, and social propagation behavior more.

From this perspective, Galxe's moat isn't in the Quest page itself, as task pages, reward mechanisms, and NFT badges can be copied. What's truly difficult to replicate is the long-term accumulation of user identity data and behavioral networks. As more and more projects launch activities on Galxe, users' behavioral resumes become increasingly complete. As more and more users deposit their participation records on Galxe, project teams become more willing to use Galxe to filter target users. Ultimately, a mutually reinforcing growth relationship forms between the platform, projects, and users: more projects lead to richer behavioral data; richer data leads to more precise user filtering; more precise filtering leads to greater project reliance on the platform.

Gamified Growth Paths and Ecosystem Synergy

Another key capability of Galxe is that it doesn't design growth as a simple "complete task → claim reward" flow but reorganizes originally fragmented growth actions into a continuous behavioral system. Most Web3 projects, when pursuing growth, often fall into two extremes: either the barrier is too high, immediately requiring users to connect wallets, perform cross-chain transactions, trade, or provide liquidity; or the barrier is too low, only involving lightweight behaviors like follows, retweets, or community joins, ultimately failing to drive real product usage.

Galxe's cleverness lies in dissecting these behaviors into a gradually escalating task ladder, allowing users to transition from "observer" to "participant" to "ecosystem user" almost unconsciously.

This path typically begins with near-zero-cost social actions. For example, following official accounts, retweeting content, joining Discord, browsing project pages. The purpose of these tasks isn't to prove user quality but to first lower the psychological barrier for initial participation, expanding the reach of the activity. Once users complete these initial low-cost actions, Galxe can then use subsequent tasks to guide them to connect their wallets, claim NFTs, complete identity verification, or visit specified dApps. The goal at this stage is to shift users from Web2-style observation to Web3-style participation, converting social traffic into identifiable wallet users.

After users complete wallet connection and basic on-chain operations, tasks escalate to higher-value on-chain behaviors, such as cross-chain bridging, swaps, minting, lending, voting, staking, or using ecosystem applications. These behaviors are what truly matter to project teams, as they indicate not only user awareness but also willingness to invest time, pay gas costs, and take some operational risk. Galxe breaks down these complex actions into smaller, achievable goals through task sequences, providing users with feedback and rewards for each step, thereby reducing the psychological resistance associated with complex on-chain operations.

In a sense, Galxe is more like reorganizing growth behavior using gamification mechanics. Users aren't abruptly pushed toward high-barrier operations; instead, while continuously completing tasks, receiving feedback, and accumulating achievements, they gradually enter deeper ecosystem participation. This is also why Galxe's growth model often shows pronounced effectiveness in large-scale ecosystem campaigns.

Taking L2 or new L1 chain ecosystems as an example, the hardest part for an ecosystem isn't getting users to "know about it," but getting them to truly experience multiple applications within it. If relying solely on individual project promotion, users might only reach a cognitive level. However, through Galxe's task system, an ecosystem can package multiple applications into an exploration roadmap, guiding users to experience different modules like wallets, bridges, DEXs, NFT marketplaces, games, and social apps in a sequential order. This way, growth is no longer point-based user acquisition but becomes an organized ecosystem tour. While completing tasks, users effectively undergo ecosystem education, product trials, and behavioral sedimentation, while project teams simultaneously gain traffic, interaction data, and potential user screening.

At a deeper level, Galxe's task system also addresses the problem of "incentive-behavior misalignment" in Web3 growth. Many projects, when distributing rewards, can only broadly incentivize a single outcome, such as making one trade, minting one NFT, or joining a community. This easily attracts many low-quality users. Galxe's approach is to break down the outcome into a process, design the process into a path, and then match different rewards to different tiered behaviors. Low-barrier tasks offer light rewards; high-value tasks offer more scarce benefits; continuous task completion grants higher-level qualifications or identity credentials. This way, user quality is gradually screened during the task process: those only willing to retweet stay at the shallow level; those willing to connect wallets move to the intermediate level; and those willing to persistently interact and complete complex tasks become higher-value users.

Therefore, what Galxe does isn't just campaign operations; it's redesigning the participation path for Web3 users. It transforms the originally chaotic growth flow into a gamified system with entry points, progression, feedback, and screening. What users experience is task completion and reward acquisition, while what project teams gain is user education, behavior guidance, data sedimentation, and user segmentation.

The Data Flywheel and Platformization Strategy

As the product continuously evolves, Galxe is no longer content with just the Quest platform positioning. It has gradually introduced products like Passport, Starboard, Earndrop, and Gravity, aiming to cover the entire Web3 growth chain: Quest handles user behavior guidance, Passport handles identity verification, Starboard handles community data analytics and contributor identification, Earndrop handles reward distribution, and Gravity extends further into underlying infrastructure.

This means Galxe is transitioning from a task tool into a complete growth operating system.

The truly difficult-to-replicate aspect isn't the task page itself, but the data network and ecosystem network it is gradually forming. As more and more projects integrate, Galxe can accumulate increasingly rich user behavioral data, helping projects filter more precise user groups. As more and more users deposit their identities and historical behaviors, the user profiles on the platform become increasingly complete.

Ultimately, Galxe forms a classic platform flywheel: more projects lead to more users; more users lead to richer behavioral data; richer data leads to more precise user screening; more precise screening makes project teams more willing to continue investing growth resources on the platform.

In a sense, what Galxe aims to be is not the largest task platform in Web3, but more akin to Google Ads in the Web3 world—it isn't truly operating tasks, but rather the growth network built around identity, behavior, and distribution.

Conclusion

If the growth of Web3 in the past essentially remained in the "traffic-thinking" stage, then the emergence of Galxe signifies the industry's first genuine attempt to establish "identity-thinking." Over the past few years, numerous projects relied on airdrops, communities, and token incentives for cold starts, but the problems with this model are equally evident: users come for the rewards and leave when the rewards end; what projects gain is often short-term data, not long-term relationships.

What Galxe truly changes is that it begins to give user behavior continuous, accumulating value. A wallet is no longer just a one-time interaction tool but gradually becomes a long-term account with historical records, participation histories, and identity credibility. Which ecosystems a user has participated in, what behaviors they've completed, and whether they've been consistently active—all these gradually sediment into a verifiable, accumulable identity asset.

This is also why Galxe's value doesn't lie solely in Quest, NFTs, or airdrops themselves, but in its push to shift Web3's growth logic from "reward-driven" toward "identity-driven." As more and more projects begin designing growth around users' historical behaviors, and as more and more users start valuing their on-chain resumes over just short-term gains, the growth approach of Web3 will become entirely different from the past. Many see a task platform, but Galxe is more like building a new growth order: user behaviors are recorded long-term, identity value is accumulated continuously, and growth is no longer just one-off traffic transactions but gradually becomes a long-term relationship network built around identity.

Domande pertinenti

QHow has Galxe evolved from a typical Quest platform to become a growth infrastructure in Web3?

AGalxe has evolved from a typical Quest platform to Web3 growth infrastructure by productizing, systematizing, and data-fying the previously fragmented, short-cycle, and non-reusable growth processes. It transforms isolated user activities into long-term, verifiable identity data through mechanisms like Credentials, OATs, Passport, and Score. This creates a platform where user behavior accumulates value over time, enabling projects to target users based on their historical on-chain and off-chain activities. By building a data network and ecosystem around identity and growth, Galxe functions as a foundational layer for user acquisition, screening, and engagement in Web3.

QWhat core growth problem in Web3 does Galxe aim to solve, and what was the inspiration behind its approach?

AGalxe aims to solve the core growth problem in Web3: the lack of mature user growth infrastructure akin to Web2 systems like Facebook Ads or Google Ads. Web3 projects struggle to identify real users, lack unified identity systems, and rely on fragmented methods like airdrops and community building, leading to short-term engagement rather than long-term user retention. The inspiration behind Galxe's approach comes from its founders' background in Web2 products like the live-streaming platform DLive, which relied heavily on community incentives and user growth. They applied proven Web2 growth logic—such as gamification, tiered task structures, and continuous feedback—to Web3, focusing on recording, verifying, and reusing user behavior.

QExplain the mechanism of 'user behavior assetization' in Galxe and its benefits for both users and projects.

AGalxe's 'user behavior assetization' mechanism converts scattered, short-term user actions into long-term, verifiable identity data recorded as Credentials, OATs, Passport, or Score. For users, this means their activities (e.g., participating in Quests, on-chain interactions) are no longer one-time actions but become part of a cumulative identity profile. This encourages 'account cultivation,' as a richer history increases the likelihood of future rewards and access. For projects, Galxe provides a filtered user pool tagged with historical behavior, allowing them to target specific user segments—like DeFi users, testnet participants, or NFT collectors—based on verified data, leading to more efficient and higher-quality growth campaigns.

QDescribe how Galxe uses gamification to design user growth paths and address the mismatch between incentives and user behavior in Web3.

AGalxe uses gamification to design a tiered growth path that guides users from low-cost social actions (e.g., following Twitter, joining Discord) to higher-value on-chain behaviors (e.g., swaps, minting, staking). This step-by-step approach lowers the psychological barrier to entry and provides continuous feedback and rewards. It addresses the Web3 incentive-behavior mismatch by aligning rewards with task complexity: lightweight tasks offer minor rewards, while complex tasks grant scarcer benefits or credentials. This naturally filters user quality, as only motivated users progress to high-value actions, ensuring projects attract engaged participants rather than mere airdrop hunters.

QWhat is the 'data flywheel' effect in Galxe's platform strategy, and how does it contribute to its defensibility as a growth infrastructure?

AThe 'data flywheel' effect in Galxe refers to a self-reinforcing cycle where more projects on the platform attract more users, whose accumulated behavior data enriches the user profiles, enabling better targeting for projects, which in turn attracts more projects to use Galxe. This network effect creates defensibility, as Galxe's value lies not in its Quest interface—which is easily replicable—but in the vast, interconnected ecosystem of user identity data and project relationships. By expanding into products like Passport (identity verification), Starboard (community analytics), and Earndrop (reward distribution), Galxe solidifies its role as a comprehensive growth operating system, making it difficult for competitors to replicate its data network and ecosystem synergy.

Letture associate

Three Years Later: Looking Back at My Predictions About ChatGPT in 2023

Three Years Later: Revisiting My 2023 Predictions on ChatGPT In March 2023, shortly after ChatGPT's launch, I made 20 predictions about its future. Now, in mid-2026, I've used AI agents to fact-check each one against the latest data. Overall, most major directional forecasts were correct, with only one outright error (incorrectly stating GPT-4 had 100 trillion parameters). Key successes included predicting that RAG and retrieval architectures would become the standard for handling knowledge and hallucinations, that natural language interfaces (LUI) would create a massive new industry layer beyond the models themselves, and that China would develop viable large language models, significantly closing the performance gap with Western counterparts within about three years. Predictions about the absence of mass unemployment, the rise of a new "robot network" for agent communication, and ChatGPT not possessing consciousness also held true in their core arguments. However, the "devil was in the details." Errors frequently involved specific numbers, timelines, or overlooking distributional effects. I tended to overestimate the speed of adoption (e.g., for agent networks) while underestimating the ultimate scale of capabilities or costs (e.g., AI winning IMO gold without tools, or the extreme capital required for frontier models). Other misjudgments included: underestimating how AI would reinforce, not dissolve, information filter bubbles; incorrectly assuming AI-generated content would easily circumvent copyright (it has instead triggered record-breaking settlements); and misidentifying where value would be captured (it accrued overwhelmingly to the compute layer, like Nvidia, not just the application or model layers). Key lessons from reviewing these predictions are: 1) Directional and mechanistic insights are far more reliable than precise numbers or absolute statements. 2) There's a consistent bias to overestimate short-term speed but underestimate long-term magnitude. 3) Errors often lie in missing distributional impacts within a generally correct aggregate trend. 4) Predictions phrased with nuance and caveats aged the best. 5) Some fundamental debates (e.g., on machine consciousness or the ultimate value chain) remain unresolved even after three years. This exercise is less about scoring the past and more about establishing rules for clearer thinking about the next three years of AI.

marsbit5 h fa

Three Years Later: Looking Back at My Predictions About ChatGPT in 2023

marsbit5 h fa

Three Years Later: Looking Back on My 2023 Predictions for ChatGPT

Looking Back After Three Years: Revisiting My 2023 Predictions on ChatGPT In March 2023, shortly after ChatGPT's debut and before GPT-4's release, I made over twenty predictions about AI's future based on limited information and intuition. Now, in May 2026, I revisited those forecasts using an AI-driven analysis with 41 Opus 4.8 agents to cross-reference them with the latest data. The assessment used symbols: ✅ Correct, 🟢 Mostly Correct, 🟡 Partially Correct, ❌ Incorrect. Overall, the directional judgments held up well, with only one major factual error regarding GPT-4's rumored parameter size (incorrectly cited as 100T). However, nuances and degrees of accuracy revealed more. **What Was Largely Correct:** Predictions about mechanisms and directions proved accurate. The rise of RAG (Retrieval-Augmented Generation) as the standard architecture for combating AI hallucination was confirmed, as was the transformative potential of LUI (Language User Interface) in creating a new industry layer atop GUIs. The emergence of "robot networks" (agent-to-agent communication protocols) and China's rapid catch-up in developing capable large models (closing the performance gap with top models to ~2.7%) were also on point. The analysis affirmed that LLMs lack consciousness and that the Turing Test merely measures perceived intelligence. **What Was Off Target:** Errors often involved specific numbers, over-optimistic timelines, or misjudged distributions. The prediction that value would primarily accrue to the application layer was half-right but missed NVIDIA's dominance as the profitable infrastructure layer. Forecasts about AI circumventing copyright issues and fostering a "global common ground" by averaging human viewpoints were incorrect; instead, major copyright settlements occurred and AI personalization is increasing. Estimates for model training costs ("$5-10 billion cap") were significantly off, underestimating frontier costs and overestimating replication costs. The notion that LLMs could never do complex math without tools was disproven by later models winning IMO gold. **Key Patterns from the Review:** 1. **Direction over precision:** Judgments about mechanisms and trends were more reliable than specific numbers or definitive statements. 2. **Timing bias:** There was a tendency to overestimate short-term speed but underestimate long-term magnitude and transformation. 3. **The distribution blind spot:** Aggregate-level correctness often masked uneven impacts (e.g., on young professionals' employment). 4. **The value of qualifiers:** Predictions framed with caution (e.g., "reportedly," "for now," "prototype in 2-3 years") aged better. 5. **Some debates continue:** Issues like the nature of "emergent abilities" or machine consciousness remain unresolved. This three-year review highlights that while seeing the big picture is crucial, humility regarding specifics, timelines, and disparate impacts is essential for future forecasting.

链捕手7 h fa

Three Years Later: Looking Back on My 2023 Predictions for ChatGPT

链捕手7 h fa

AI Bubble Warning: AI Investments Are Negative Returns for Most Tech Giants

The article issues a stark warning about a potential AI investment bubble. It notes that while the AI boom shares similarities with the TMT bubble of the late 1990s, its scale is vastly larger, currently driving 93% of U.S. GDP growth. Major hyperscale cloud providers like Microsoft, Alphabet, Amazon, Meta, and Oracle are planning to invest trillions in AI data centers over the coming years. However, calculations based on analyst projections for 2025-2030 reveal a concerning math problem: expected capital expenditure growth far outpaces projected revenue growth. Even under an extremely optimistic scenario of zero costs, the implied return on investment for most of these tech giants (except Amazon) is deeply negative. This suggests that the current trajectory could lead to one of history's largest shareholder value destruction events. The piece outlines two potential escapes: AI generating vastly more revenue than currently anticipated—a near-impossible task—or a significant cutback in the planned investment splurge. The latter scenario could trigger a domino effect, severely impacting the entire tech supply chain (from Nvidia to TSMC), potentially pushing the U.S. economy into recession, and causing a major stock market downturn. The author suggests upcoming high-profile IPOs by companies like OpenAI and Anthropic might represent a transfer of risk from early investors to public market participants. While the peak of the hype cycle might sustain investment through 2026, the fundamental financial dilemma remains unresolved, setting the stage for a potential market correction in 2027 or 2028, similar to the years following Alan Greenspan's "irrational exuberance" warning.

marsbit8 h fa

AI Bubble Warning: AI Investments Are Negative Returns for Most Tech Giants

marsbit8 h fa

From Tokens to Machine Labor: AI is Shifting from Tool to "Worker"

The article "From Token to Machine Labor: AI is Evolving from Tool to 'Worker'" argues that the business model for AI is shifting beyond simply selling computational resources (tokens, GPU hours) or model access. Instead, a new "machine labor market" is emerging, where the core economic transaction is the purchase of economically useful work directly performed by software. The central thesis is that AI pricing will evolve through four stages: 1) raw tokens, 2) standardized LLM capabilities (e.g., text generation), 3) industry-specific labor markets (e.g., legal review, radiology), and finally 4) a programmable results market where tasks like resolving a support ticket are bid on and priced based on outcome. In this future, buyers will care less about *which* model or GPU completes a task and more about whether the work meets specified standards for accuracy, latency, and cost. This transition reframes the impact of AI on human labor. Rather than simple replacement, it suggests a re-coordination where machines handle standardized, verifiable work, freeing humans for roles involving oversight, context management, responsibility, and final judgment. In some cases, this "last 1%" of human input becomes more valuable as it enables the other 99% to be automated. Furthermore, as AI reduces the cost of work, demand may expand, creating larger markets (e.g., 24/7 customer service) rather than just cheaper versions of existing ones. The article concludes that while infrastructure (GPUs, models, tokens) remains crucial upstream, the market is converging on a simpler, tradeable unit: machine labor that can be defined, measured, priced, and procured based on contractible specifications.

marsbit8 h fa

From Tokens to Machine Labor: AI is Shifting from Tool to "Worker"

marsbit8 h fa

Xiaomi MiMo's 99% Price Cut is Not Marketing! Luo Fuli Posts on X to Refute Critics

The price of Xiaomi's MiMo-V2.5 series API has been permanently reduced by up to 99%, specifically for the "Input (Cache Hit)" cost, which covers users re-reading historical context in long conversations. MiMo's head, Luo Fuli, published a detailed technical blog to clarify that this drastic price cut stems from genuine engineering breakthroughs, not a marketing stunt or a simple price war. The core of the achievement lies in six key engineering optimizations. First, the model architecture adopts a Hybrid Sliding Window Attention (SWA), reducing the memory footprint (KVCache) to 1/7th of a traditional model. Second, a dual-pool memory management system actually utilizes these savings, allowing a single GPU to handle over 5 times more concurrent users. Third, an upgraded prefix caching mechanism achieves a cache hit rate of 93-95% for repeated reads, meaning most such requests bypass GPU computation entirely. Fourth, a self-developed distributed cache (GCache) utilizes idle SSD space on existing GPU servers, eliminating additional storage costs. Fifth, an intelligent scheduling system (LLM-Router) efficiently routes requests to maximize cache reuse and performance. Sixth, Multi-Token Prediction (MTP) accelerates the model's text generation ("output") side. Together, these systemic optimizations dramatically lower the real computational cost per request, enabling the 99% price reduction for cached inputs while reportedly maintaining positive gross margins. Luo Fuli's disclosure aims to shift the narrative from "price war" to a demonstration of substantive AI engineering progress.

marsbit10 h fa

Xiaomi MiMo's 99% Price Cut is Not Marketing! Luo Fuli Posts on X to Refute Critics

marsbit10 h fa

Trading

Spot
Futures

Articoli Popolari

Come comprare PEOPLE

Benvenuto in HTX.com! Abbiamo reso l'acquisto di ConstitutionDAO (PEOPLE) semplice e conveniente. Segui la nostra guida passo passo per intraprendere il tuo viaggio nel mondo delle criptovalute.Step 1: Crea il tuo Account HTXUsa la tua email o numero di telefono per registrarti il tuo account gratuito su HTX. Vivi un'esperienza facile e sblocca tutte le funzionalità,Crea il mio accountStep 2: Vai in Acquista crypto e seleziona il tuo metodo di pagamentoCarta di credito/debito: utilizza la tua Visa o Mastercard per acquistare immediatamente ConstitutionDAOPEOPLE.Bilancio: Usa i fondi dal bilancio del tuo account HTX per fare trading senza problemi.Terze parti: abbiamo aggiunto metodi di pagamento molto utilizzati come Google Pay e Apple Pay per maggiore comodità.P2P: Fai trading direttamente con altri utenti HTX.Over-the-Counter (OTC): Offriamo servizi su misura e tassi di cambio competitivi per i trader.Step 3: Conserva ConstitutionDAO (PEOPLE)Dopo aver acquistato ConstitutionDAO (PEOPLE), conserva nel tuo account HTX. In alternativa, puoi inviare tramite trasferimento blockchain o scambiare per altre criptovalute.Step 4: Scambia ConstitutionDAO (PEOPLE)Scambia facilmente ConstitutionDAO (PEOPLE) nel mercato spot di HTX. Accedi al tuo account, seleziona la tua coppia di trading, esegui le tue operazioni e monitora in tempo reale. Offriamo un'esperienza user-friendly sia per chi ha appena iniziato che per i trader più esperti.

438 Totale visualizzazioniPubblicato il 2024.12.12Aggiornato il 2025.03.21

Come comprare PEOPLE

Discussioni

Benvenuto nella Community HTX. Qui puoi rimanere informato sugli ultimi sviluppi della piattaforma e accedere ad approfondimenti esperti sul mercato. Le opinioni degli utenti sul prezzo di PEOPLE PEOPLE sono presentate come di seguito.

活动图片