Model updates

Official news, first rumor reports and substantive progress. Each entry keeps the information published at that time.

• Zhipu quietly tests GLM-5.3-Highspeed in Zcode, with a user clocking 490 tokens/s, per QbitAI

  QbitAI reports that Zhipu is rolling out a GLM-5.3-Highspeed model to random subscribers in Zcode 3.15.1, where one user showed output at 490 tokens per second, versus a 200 tokens/s peak for last month's GLM-5.3-FlashX. Pi's model catalog lists it with a 1M-token context and 128k max output; Zhipu has not formally announced it.

  Reported by QbitAIPublished newsOct 11Source: QbitAIPart of GLM
• Zhipu's GLM-5.3 arrives on Amazon Bedrock for enterprise customers

  Z.ai's GLM-5.3 is now available on Amazon Bedrock, with enterprise customers able to call it through Bedrock's managed API without deploying their own inference infrastructure, per multiple reports. AWS pays Zhipu per invocation, the first time a Chinese frontier lab's model earns usage-based revenue on AWS. Access requires an AWS account and IAM permissions via the standard Bedrock API.

  UnconfirmedPublished newsOct 6Source: X @0xLogicrwPart of GLM
• 700B-parameter GLM runs on a laptop with SSD as memory, per QbitAI

  QbitAI reports Colibri, a fast-growing open-source project on GitHub, runs 700-billion-parameter GLM models on laptops without a GPU by using SSD storage as memory.

  Reported by QbitAIPublished newsSep 26Source: QbitAIPart of GLM

RSS for this view →