The New Stack says Flash took 7.6 minutes and 38,677 tokens on the hardest task, while GLM-5.3 reached the same answer in under three minutes.
The New Stack reports that Z.AI launched GLM-5.3-Flash on August 26, promising flagship-grade intelligence at an exceptionally low cost. In the publisher’s comparison with GLM-5.3, the models earned identical scores. But Flash took 7.6 minutes and used 38,677 tokens on the hardest task; the flagship reached the answer in under three minutes.
572
1 post, first seen 6d ago
The New Stack says Flash took 7.6 minutes and 38,677 tokens on the hardest task, while GLM-5.3 reached the same answer in under three minutes.
The New Stack reports that Z.AI launched GLM-5.3-Flash on August 26, promising flagship-grade intelligence at an exceptionally low cost. In the publisher’s comparison with GLM-5.3, the models earned identical scores. But Flash took 7.6 minutes and used 38,677 tokens on the hardest task; the flagship reached the answer in under three minutes.
The New Stack@thenewstackGLM-5.3-Flash vs. GLM-5.3: Time and money, not the spec sheet
The New Stack@thenewstackGLM-5.3-Flash vs. GLM-5.3: Time and money, not the spec sheet
