Básico
Spot
Opera con criptomonedas libremente
Margen
Multiplica tus beneficios con el apalancamiento
Convertir e Inversión automática
0 Fees
Opera cualquier volumen sin tarifas ni deslizamiento
ETF
Obtén exposición a posiciones apalancadas de forma sencilla
Trading premercado
Opera nuevos tokens antes de su listado
Contrato
Accede a cientos de contratos perpetuos
TradFi
Oro
Plataforma global de activos tradicionales
Opciones
Hot
Opera con opciones estándar al estilo europeo
Cuenta unificada
Maximiza la eficacia de tu capital
Trading de prueba
Introducción al trading de futuros
Prepárate para operar con futuros
Eventos de futuros
Únete a eventos para ganar recompensas
Trading de prueba
Usa fondos virtuales para probar el trading sin asumir riesgos
Lanzamiento
CandyDrop
Acumula golosinas para ganar airdrops
Launchpool
Staking rápido, ¡gana nuevos tokens con potencial!
HODLer Airdrop
Holdea GT y consigue airdrops enormes gratis
Launchpad
Anticípate a los demás en el próximo gran proyecto de tokens
Puntos Alpha
Opera activos on-chain y recibe airdrops
Puntos de futuros
Gana puntos de futuros y reclama recompensas de airdrop
Inversión
Simple Earn
Genera intereses con los tokens inactivos
Inversión automática
Invierte automáticamente de forma regular
Inversión dual
Aprovecha la volatilidad del mercado
Staking flexible
Gana recompensas con el staking flexible
Préstamo de criptomonedas
0 Fees
Usa tu cripto como garantía y pide otra en préstamo
Centro de préstamos
Centro de préstamos integral
Centro de patrimonio VIP
Planes de aumento patrimonial prémium
Gestión patrimonial privada
Asignación de activos prémium
Quant Fund
Estrategias cuantitativas de alto nivel
Staking
Haz staking de criptomonedas para ganar en productos PoS
Apalancamiento inteligente
Apalancamiento sin liquidación
Acuñación de GUSD
Acuña GUSD y gana rentabilidad de RWA
智譜: La rentabilidad bruta de la plataforma MaaS aumenta casi 5 veces
Spire (02513) The chief executive officer, Zhang Peng, disclosed in the results briefing that the company’s MaaS platform (Model as a Service, namely open platform and API business) ARR (annual recurring revenue) has reached approximately RMB 1.7 billion as of March. In the past 12 months, it has increased 60-fold. The MaaS platform’s gross margin has increased nearly 5-fold to 18.9%.
He also said that as of this March, even though the API calling pricing of the MaaS platform was increased by 83% compared with the end of last year, demand still outstrips supply, indicating that high-end intelligence is a scarce resource at present. “Whoever controls the intelligence upper bound (top-tier model capability), whoever controls pricing power.”
Zhang Peng said that today AI capabilities have moved from “available” and “playable” to real production, solving increasingly complex problems, so that API calls and token consumption can truly translate into economic value. Combined with new application forms and device-level native intelligence development such as OpenClaw, token consumption in the future can be expected to show exponential growth.
Regarding the company’s positioning, Zhang Peng said that over the past year, Anthropic has been one of the most closely watched AI companies globally. Its growth logic is clear: it focuses on delivering the strongest models to enterprises and developers via APIs. Such a business model and growth are also happening at Spire: “When the model is strong enough, the API itself is the best business model.” In 2026, Spire will continue to develop along the business path of “the Anthropic of China,” with model intelligence as the foundation and the API platform as the engine.
When asked about the impact of large models on the traditional software industry, Zhang Peng believes that the software and IT (information technology) industries will see a new paradigm emerge. Some traditional software companies may be affected—this is a normal phenomenon of the replacement brought about by a new era.
As for compute power shortages, Spire stated that facing the shortage in compute power supply that has seen demand outstrip supply since February this year, it will continue to increase investment, especially by deepening localization chip compatibility and software-hardware integrated optimization, to “squeeze” inference performance to the limit.