Unverified50% confidenceEventExact time
A Bilibili uploader used llama.cpp to load the 1-bit quantized version of GLM 5.3 Flash on 4 RTX 3090s, achieving a generation speed of 33.5 tok/s.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/15/2026
First Seen
Valid until: 10/29/2026
Sources
GLM 5.3 Flash Local Test: Fitting a 320B Model into 4 RTX 3090s with 1-bit Quantization
bilibili四月的尾巴_10/13/2026
Related Entities
Related Claims
UnverifiedZhipu has released GLM-4.5 FlashX, reaching a maximum speed of 200 tokens/second, with the API now live.75% similarUnverifiedOn an RTX 3060, the developer's hand-written Triton fused W4A16 GEMV kernel targeting batch-1 decoding runs at about 25.5 tok/s, matching the roughly 25.7 tok/s of the bitsandbytes NF4 quantization scheme.73% similarUnverifiedLTX-2.3 video clip generation speed has increased by approximately 4.5×, thanks to distilled sampling, compilation fixes, and hosted FP8 weights.70% similarUnverifiedMiMo V2.6 Flash has 310B total parameters and 15B activated parameters, with a speed nearly twice that of Pro.70% similarUnverifiedThe previous generation v4 Flash was released in July with 284 billion parameters, and the community can pack the 4-bit experts into a 128 GB gguf version that can be run on four second-hand RTX 3060s.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1035997API
curl https://kongchang.com/api/v1/knowledge/claims/1035997MCP
get_claim(id=1035997)