Z.ai released GLM-5.3-Flash on August 26, 2026, shipping it as a 320-billion-parameter mixture-of-experts model that activates just 18 billion parameters per token, with MIT-licensed weights posted to Hugging Face on day one. According to Z.ai's announcement quoted in the piece, it is the first natively multimodal mode API pricing for GLM-5.3-Flash is set at $0.15 per million input tokens and $0.50 per million output tokens, roughly one-ninth the cost of the GLM-5.3 flagship's $1.40/$4.40 rates. The piece also notes that Z.ai's promised open-weight release of the GLM-5.3 flagship's 744B parameters, expected around launch, has not mat

A side-by-side comparison between GLM-5.3 and GLM-5.3-Flash draws on Artificial Analysis testing and Z.ai's published specs. GLM-5.3 scores 60 on the Intelligence Index versus 57 for GLM-5.3-Flash, but GLM-5.3-Flash activates only 18B parameters (vs. 40B) and costs roughly one-ninth as much at normal API pricing ($0.15 A notable counter-narrative finding: GLM-5.3-Flash is not faster at generation than the flagship. Artificial Analysis measures output at roughly 49 tokens per second for Flash versus 85 tok/s for GLM-5.3, with time-to-first-token nearly identical (1.46s vs. 1.57s). The article concludes GLM-5.3-Flash is the better valu

GLM-5.3-Flash shipped on August 26, 2026 as a 320B-parameter mixture-of-experts model activating 18B parameters per token, trained natively in FP8 with a 1,048,576-token context window and described as the first natively multimodal model in the GLM-5 series. The deep-dive confirms it is built from a newly trained base Benchmark coverage shows GLM-5.3-Flash beating GLM-5.2 across six coding and agentic benchmarks at roughly one-tenth the price, and approaching Claude Opus 4.8 on long-horizon agent tasks. The article surfaces developer-facing details: MIT-licensed weights on Hugging Face, API pricing at $0.15/$0.50 per million tokens,