مرّر فوق النص للترجمة · Hover text to translate · Tap on mobile
GLM-5.3-Flash
Frontier coding intelligence at flash cost - 320B open model from $0.15/1M tokensذكاء برمجي بمستوى الطليعة بسعر خاطف - نموذج مفتوح 320B من 0.15$ للمليون رمزFrontier coding intelligence at flash cost - 320B open model from $0.15/1M tokens
Z.ai's GLM-5.3-Flash (Aug 2026) is a 320B-parameter natively multimodal MoE (18B active) with a 1M-token context and MIT-licensed weights on Hugging Face. It beats GLM-5.2 across benchmarks at one-tenth the price and nears Claude Opus 4.8 on coding (84.3 Terminal-Bench 2.1, 63.4 DeepSWE) - scoring 57 on the Artificial Analysis Index at just $0.045/task. Note: "Flash" is not a distilled mini, it is a newly trained base. API from $0.15/1M input tokens, or 3x quota on the GLM Coding Plan; it quietly topped OpenRouter charts pre-launch as the anonymous "Ox Alpha".GLM-5.3-Flash من Z.ai (أغسطس 2026) نموذج MoE متعدد الوسائط أصيلاً بـ 320 مليار معامل (18B نشطة) وسياق مليون رمز، بأوزان مفتوحة برخصة MIT على Hugging Face. يتفوق على GLM-5.2 بعُشر السعر ويقارب Claude Opus 4.8 في البرمجة. الـ API من 0.15$ للمليون رمز مع 3 أضعاف الحصة على خطة GLM Coding. انتبه: "Flash" ليس نسخة مصغرة بل أساس جديد مدرّب بالكامل.Z.ai's GLM-5.3-Flash (Aug 2026) is a 320B-parameter natively multimodal MoE (18B active) with a 1M-token context and MIT-licensed weights on Hugging Face. It beats GLM-5.2 across benchmarks at one-tenth the price and nears Claude Opus 4.8 on coding (84.3 Terminal-Bench 2.1, 63.4 DeepSWE) - scoring 57 on the Artificial Analysis Index at just $0.045/task. Note: "Flash" is not a distilled mini, it is a newly trained base. API from $0.15/1M input tokens, or 3x quota on the GLM Coding Plan; it quietly topped OpenRouter charts pre-launch as the anonymous "Ox Alpha".
Read the launch notesاقرأ إعلان الإطلاقRead the launch notes
Skim the Z.ai announcement for benchmarks, pricing, and the Ox Alpha backstory.اطّلع على المعايير والأسعار وقصة Ox Alpha.Skim the Z.ai announcement for benchmarks, pricing, and the Ox Alpha backstory.
2
Grab the open weights (optional)حمّل الأوزان المفتوحة (اختياري)Grab the open weights (optional)
Pull zai-org/GLM-5.3-Flash from Hugging Face (MIT license) to self-host with SGLang, vLLM, or TokenSpeed - or skip this and use the hosted API.اسحب النموذج من Hugging Face برخصة MIT للاستضافة الذاتية.Pull zai-org/GLM-5.3-Flash from Hugging Face (MIT license) to self-host with SGLang, vLLM, or TokenSpeed - or skip this and use the hosted API.
3
Get API accessاحصل على وصول APIاحصل على وصول API
Create a Z.ai account and generate an API key at $0.15/1M input and $0.50/1M output tokens - or subscribe to the GLM Coding Plan (Lite $18/mo) for 3x the quota of GLM-5.3.أنشئ حساب Z.ai بـ 0.15$ للمليون رمز أو اشترك في GLM Coding Plan.Create a Z.ai account and generate an API key at $0.15/1M input and $0.50/1M output tokens - or subscribe to the GLM Coding Plan (Lite $18/mo) for 3x the quota of GLM-5.3.
4
Use it in your agentاستخدمه في وكيلكUse it in your agent
Point any OpenAI-compatible client at the Z.ai endpoint with model code glm-5.3-flash - it also serves via OpenRouter as z-ai/glm-5.3-flash.وجّه أي عميل متوافق مع OpenAI إلى glm-5.3-flash.Point any OpenAI-compatible client at the Z.ai endpoint with model code glm-5.3-flash - it also serves via OpenRouter as z-ai/glm-5.3-flash.
