Model Spend Arena2ND ED.
builds models · serves endpoints · 2026-10-10

Z.ai / Zhipu

Official site · 🇨🇳 China

Plan

GLM Coding (Lite / Pro / Max)

Tier$/monthIncluded usage
Lite$18.00—
Pro$80.00—
Max$168.00—

console only quota converts to units

As advertised. 2.000 / 12.000 / 28.000 creditos por 5h; 10.000 / 60.000 / 140.000 por semana (confirmado en docs.z.ai/devpack/overview el 2026-09-06). La pagina de venta ya no ensena creditos: dice '6x Lite' y '14x Lite'.

What that actually means. Publican la formula entera, asi que los creditos SI convierten a tokens: creditos = (tokens de entrada x mult_in + cacheados x mult_cache + salida x mult_out) / 10.000. Multiplicadores por modelo: GLM-5.3 6,9 / 1,7 / 24; GLM-5.3-Flash 2,3 / 0,56 / 8; las llamadas MCP (Web Search, Web Reader, Zread) 1,2 de salida cada una. Y hay tabla de tokens por semana segun el acierto de cache — con GLM-5.3 al 95%, Lite 48-97 M, Pro 290-580 M y Max 676-1.352 M. La propia pagina dice que ahorra hasta un 92% contra llamar a la API por tokens.

Where to check what is left. z.ai subscription page; an undocumented monitor endpoint also exists and was verified directly

Models covered. GLM-5.3, GLM-5.3-Flash

CORREGIDO 2026-09-11, y la ficha anterior decia lo contrario de lo que pasa. El esquema de prompts con recargo en hora punta MURIO el 2026-07-30. Hoy se mide en creditos sobre tokens, y lo que hay no es recargo sino DESCUENTO: fuera de punta se cobra el 50% de la tarifa. La punta son lunes a viernes de 14:00 a 18:00 hora de Singapur (20 horas a la semana), y el fin de semana entero va a tarifa de fuera de punta. O sea que la penalizacion real es mucho menor de lo que decia esta ficha, y va en sentido contrario. 2026-09-24: la promo de GLM-5.3-Flash ya tiene ventana publicada: del 3 de septiembre al 7 de octubre, cada dia de 23:00 a 09:00, uso ilimitado por ZCode y AutoClaw para los planes de pago. La pagina ensena ademas trimestral con 20% y anual con 30% de descuento. 2026-10-02: promo nueva, 'From September 25 to October 7, 2026, all-day usage will be charged at the off-peak rate': hasta el 7 de octubre la punta no cobra recargo, o sea 50% de creditos a cualquier hora. La campana de GLM-5.3-Flash (3 sep a 7 oct, 23:00-09:00, ilimitado por ZCode y AutoClaw) seguia viva entonces. La tabla de tokens por semana se amplia con GLM-5.3-Flash (Lite 146-292, Pro 877-1.755, Max 2.047-4.095 millones por semana al 95% de acierto de cache) y con filas al 96% y 98%. Referidos nuevos, hasta el 20% de vuelta. 2026-10-09: las dos promos terminaron el 2026-10-07, como anunciaban, y ya no salen en la pagina: la punta vuelve a cobrar tarifa completa y GLM-5.3-Flash deja de ser ilimitado de noche.

checked 2026-10-09 · high (price and quota, rendered with a real browser) · source

Models it builds

ModelIntelligencetok/s$/M in$/M out$ / sessionCache
GLM-5.233.785.1$0.06$7.00$0.944790.5%
GLM-5.344.871.2$0.04$4.80——
GLM-5.3-Flash41.858.6$0.15$0.50——
GLM-5.126.154.1$0.97$3.04$0.628888.5%
GLM 4.618.534.8$0.50$2.00——
GLM 4.722.241.9$0.60$2.20——

Models it serves

6 endpoints, cheapest first. This is the view the model-by-model tables cannot give you: what one provider actually offers, and at what precision.

ModelQuant$/M in$/M outContextUptime 30m
GLM-5.3-Flashfp8$0.15$0.501,048,576100.0%
GLM 4.6fp4$0.60$2.20202,752—
GLM 4.7fp4$0.60$2.20202,752100.0%
GLM-5.2fp8$1.40$4.401,048,576100.0%
GLM-5.3fp8$1.40$4.401,048,576100.0%
GLM-5.1fp8$1.40$4.40202,752100.0%

Where to buy its models

Everyone serving these models, cheapest first, with the numeric precision each one runs. The lab is often not the cheapest place to buy its own model.

ModelServed byQuant$/M in$/M outUptime 30m
GLM-5.2Decartmxfp4$0.27$1.68100.0%
GLM-5.2StreamLakefp8$0.56$1.75100.0%
GLM-5.2DeepInfrafp4$0.56$1.80100.0%
GLM-5.2Novitafp8$0.65$2.04100.0%
GLM-5.2DigitalOceanunknown$0.70$2.20100.0%
GLM-5.2CoreWeavefp4$0.76$2.42100.0%
GLM-5.2InferenceNetfp4$0.18$4.40100.0%
GLM-5.2AtlasCloudfp8$0.94$2.95100.0%
GLM-5.2Alibaba / Qwenfp8$0.97$3.04—
GLM-5.2Morphfp8$0.18$6.00100.0%
GLM-5.2Phalafp8$1.26$3.00100.0%
GLM-5.2Alibaba / Qwenfp8$1.12$3.52100.0%
GLM-5.2Waferunknown$0.06$7.00100.0%
GLM-5.2SiliconFlowfp8$1.19$3.74100.0%
GLM-5.2Cloudflareunknown$0.50$6.00100.0%
GLM-5.2Baidufp8$1.40$4.40100.0%
GLM-5.2BaseTenfp8$1.40$4.40100.0%
GLM-5.2BaseTenfp8$1.40$4.40100.0%
GLM-5.2Friendliunknown$1.40$4.40100.0%
GLM-5.2GMICloudfp8$1.40$4.40100.0%
GLM-5.2Mistralnvfp4$1.40$4.40100.0%
GLM-5.2Nebiusfp4$1.40$4.40100.0%
GLM-5.2Parasailfp4$1.40$4.40100.0%
GLM-5.2Togetherunknown$1.40$4.40100.0%
GLM-5.2Venicefp8$1.40$4.40100.0%
GLM-5.2Z.ai / Zhipufp8$1.40$4.40100.0%
GLM-5.2Inceptronfp4$1.25$5.46100.0%
GLM-5.2Mistralunknown$1.54$4.84100.0%
GLM-5.2Waferunknown$0.19$10.00100.0%
GLM-5.2BaseTenfp8$2.10$6.60100.0%
GLM-5.2BaseTenfp8$2.10$6.60100.0%
GLM-5.2Alibaba / Qwenfp8$2.31$7.26100.0%
GLM-5.2Baidufp4$2.25$7.88—
GLM-5.2Decartfp4$2.50$9.00100.0%
GLM-5.3Inceptronfp4$0.60$2.20100.0%
GLM-5.3Sail Researchfp8$0.20$3.40100.0%
GLM-5.3Sail Researchfp8$0.20$3.40100.0%
GLM-5.3DeepInfrafp4$0.56$2.5066.7%
GLM-5.3Novitafp8$0.70$2.20100.0%
GLM-5.3Rekaunknown$0.14$4.20100.0%
GLM-5.3InferenceNetfp4$0.08$4.40100.0%
GLM-5.3Makorafp4$0.14$4.40100.0%
GLM-5.3Waferunknown$0.04$4.80100.0%
GLM-5.3AkashMLfp8$0.19$4.40100.0%
GLM-5.3Phalaunknown$0.84$2.64100.0%
GLM-5.3Decartfp4$0.84$2.65100.0%
GLM-5.3Waferunknown$0.17$4.80100.0%
GLM-5.3DigitalOceanunknown$0.91$2.86—
GLM-5.3Io Netfp8$0.75$3.40100.0%
GLM-5.3GMICloudfp8$0.98$3.08—
GLM-5.3SiliconFlowfp8$1.12$3.52—
GLM-5.3Alibaba / Qwenunknown$1.19$3.74—
GLM-5.3Friendliunknown$1.26$3.96100.0%
GLM-5.3Morphfp8$0.86$6.00100.0%
GLM-5.3AtlasCloudfp8$1.40$4.40100.0%
GLM-5.3Baidufp8$1.40$4.40100.0%
GLM-5.3BaseTenfp4$1.40$4.40100.0%
GLM-5.3BaseTenfp4$1.40$4.40100.0%
GLM-5.3Cloudflareunknown$1.40$4.40—
GLM-5.3Crusoefp4$1.40$4.40100.0%
GLM-5.3Fireworksunknown$1.40$4.40100.0%
GLM-5.3Mistralnvfp4$1.40$4.40100.0%
GLM-5.3Mistralnvfp4$1.40$4.40100.0%
GLM-5.3Modalunknown$1.40$4.40100.0%
GLM-5.3Nebiusfp4$1.40$4.4092.3%
GLM-5.3Parasailfp8$1.40$4.40100.0%
GLM-5.3PrimeIntellectunknown$1.40$4.40100.0%
GLM-5.3Togetherunknown$1.40$4.40100.0%
GLM-5.3Veniceunknown$1.40$4.40100.0%
GLM-5.3Z.ai / Zhipufp8$1.40$4.40100.0%
GLM-5.3Mistralnvfp4$1.54$4.84100.0%
GLM-5.3BaseTenfp8$2.10$6.60100.0%
GLM-5.3BaseTenfp8$2.10$6.60100.0%
GLM-5.3Fireworksunknown$2.10$6.60100.0%
GLM-5.3Alibaba / Qwenunknown$2.80$8.80—
GLM-5.3-FlashStreamLakefp8$0.07$0.23100.0%
GLM-5.3-FlashDeepInfrafp4$0.07$0.25100.0%
GLM-5.3-FlashNovitafp8$0.08$0.28100.0%
GLM-5.3-FlashGMICloudfp8$0.09$0.30100.0%
GLM-5.3-FlashOpenInferencefp4$0.04$0.45100.0%
GLM-5.3-FlashDecartfp4$0.09$0.31100.0%
GLM-5.3-FlashRelaceunknown$0.04$0.50100.0%
GLM-5.3-FlashNear AIfp8$0.10$0.35100.0%
GLM-5.3-FlashInferenceNetfp4$0.07$0.50100.0%
GLM-5.3-FlashWaferunknown$0.07$0.50100.0%
GLM-5.3-FlashPhalanvfp4$0.11$0.38100.0%
GLM-5.3-FlashSail Researchfp4$0.04$0.60100.0%
GLM-5.3-FlashSail Researchfp4$0.04$0.60100.0%
GLM-5.3-FlashInceptronfp8$0.12$0.55100.0%
GLM-5.3-FlashAtlasCloudfp8$0.15$0.507.7%
GLM-5.3-FlashBaseTenfp8$0.15$0.50100.0%
GLM-5.3-FlashBaseTenfp8$0.15$0.50100.0%
GLM-5.3-FlashCoreWeavenvfp4$0.15$0.50100.0%
GLM-5.3-FlashCrusoefp4$0.15$0.50100.0%
GLM-5.3-FlashDigitalOceanunknown$0.15$0.50100.0%
GLM-5.3-FlashFireworksunknown$0.15$0.50100.0%
GLM-5.3-FlashFriendliunknown$0.15$0.50100.0%
GLM-5.3-FlashModalnvfp4$0.15$0.50100.0%
GLM-5.3-FlashParasailfp4$0.15$0.50100.0%
GLM-5.3-FlashSiliconFlowfp8$0.15$0.5055.6%
GLM-5.3-FlashTogetherunknown$0.15$0.50100.0%
GLM-5.3-FlashVeniceunknown$0.15$0.50100.0%
GLM-5.3-FlashZ.ai / Zhipufp8$0.15$0.50100.0%
GLM-5.3-FlashMorphfp8$0.11$0.7593.3%
GLM-5.3-FlashParasailfp4$0.19$0.62100.0%
GLM-5.3-FlashDekaLLMunknown$0.10$1.00100.0%
GLM-5.3-FlashFireworksunknown$0.22$0.75100.0%
GLM-5.3-FlashRekaunknown$0.06$1.60100.0%
GLM-5.1StreamLakefp8$0.97$3.04100.0%
GLM-5.1Chutesfp8$0.98$3.08—
GLM-5.1SiliconFlowfp8$1.19$3.74100.0%
GLM-5.1AtlasCloudfp8$1.26$3.96—
GLM-5.1Phalaunknown$1.21$4.20—
GLM-5.1Alibaba / Qwenfp8$1.33$4.18—
GLM-5.1Novitafp8$1.38$4.40100.0%
GLM-5.1Baidufp8$1.40$4.40—
GLM-5.1Friendliunknown$1.40$4.40100.0%
GLM-5.1GMICloudfp8$1.40$4.40—
GLM-5.1Nebiusfp8$1.40$4.40100.0%
GLM-5.1Z.ai / Zhipufp8$1.40$4.40100.0%
GLM-5.1Venicefp8$1.40$4.40—
GLM 4.6Venicefp4$0.43$1.75100.0%
GLM 4.6DeepInfrafp4$0.50$2.00100.0%
GLM 4.6Novitabf16$0.55$2.20100.0%
GLM 4.6Z.ai / Zhipufp4$0.60$2.20—
GLM 4.7DeepInfrafp4$0.40$1.75100.0%
GLM 4.7Venicefp4$0.40$1.93100.0%
GLM 4.7Novitafp8$0.54$1.98100.0%
GLM 4.7Googleunknown$0.60$2.20100.0%
GLM 4.7Z.ai / Zhipufp4$0.60$2.20100.0%
GLM 4.7Mancer 2fp4$0.70$2.50100.0%