News for hackers. Stuff that matters.

A comment on Cerebras Code now supports GLM 4.6 at 1000 tokens/sec. Parent comment

Re:Cerebras Code now supports GLM 4.6 at 1000 tokens/sec by prodigycorp on Friday November 07, 2025 @11:06PM (#45854044)
Can you provide more context for this? (eg Was SWE-1.5 released recently? Is it considered good? Is it considered fast? Was there speculation about what the underlying model was? How does this prove that it's a GLM finetune?)
[ Parent | on HN ]
Re:Cerebras Code now supports GLM 4.6 at 1000 tokens/sec by mhuffman on Friday November 07, 2025 @11:08PM (#45854047)
I suspect they are referencing the 950tok/s claim on Cognition's page.
[ Parent | on HN ]
Re:Cerebras Code now supports GLM 4.6 at 1000 tokens/sec by prodigycorp on Friday November 07, 2025 @11:10PM (#45854052)
Ah. Thx. Blogpost for others: https://cognition.ai/blog/swe-1-5

Takeaway is that this is sonnet-ish model at 10x the speed.

[ Parent | on HN ]
Re:Cerebras Code now supports GLM 4.6 at 1000 tokens/sec by NitpickLawyer on Saturday November 08, 2025 @01:40AM (#45854647)
People saw chinese characters in generations made by swe-1.5 (windsurfs model) and also in the one made by cursor. This led to suspicions that the models are finetunes of chinese models (which makes sense, as there aren't many us/eu strong coding models out there). GLM4.5/4.6 are the "strongest" coding models atm (with dsv3.2 and qwen somewhat behind) so that's where the speculation came from. Cerebras serving them at roughly the same speeds kinda adds to that story (e.g. if it'd be something heavier like dsv3 or kimik2 it would be slower).
[ Parent | on HN ]
Re:Cerebras Code now supports GLM 4.6 at 1000 tokens/sec by prodigycorp on Saturday November 08, 2025 @02:14AM (#45854800)
Really appreciate this context. Thank you!
[ Parent | on HN ]