all the models — AI benchmark observatory
← Models

GLM-5.3-Flash

Zhipu AI · open weight · glm-5.3-flash

GLM-5.3-Flash is the first natively multimodal model in Z.AI's GLM-5 series: a 320B-total / 18B-active MoE with hybrid sparse and linear attention (plus mHC), a 1M-token context window, and strong coding and long-horizon agent performance. MIT open weights are on Hugging Face. Launched 2026-08-26 after stealth testing as ox-alpha on OpenRouter and OpenCode. List pricing is $0.15/$0.50 per million input/output tokens (a temporary 50% promo ends 2026-09-09 16:00 UTC).

GDPval-AACharXiv-RTerminal-BenToolathlonHumanity's L

Benchmark scores

BenchmarkScore
GDPval-AA1773.00
CharXiv-R0.89
Terminal-Bench 2.10.84
Toolathlon0.78
Humanity's Last Exam0.55

Pricing

  • DeepInfra$0.15 / $0.50
  • FriendliAI$0.15 / $0.50
  • Novita$0.15 / $0.50
  • ZAI$0.15 / $0.50

Input / output per 1M tokens

AA metrics

No Artificial Analysis link yet.

Arena Elo

  • vision1275
  • code1607
  • text1475