VQV Signal
According to a post on X by @bnjmn_marie, the researcher released two new SLQ quantizations of the Qwen3.8 27B model. One targets EAR 98, and the other aims to beat the official FP8 version while being about 10GB smaller. The post says GPTQ is near its limit at very low precision, and that more expensive methods are planned for Qwen4. These are the author's stated aims, and VQV has not independently tested the models.
According to a post on X by @bnjmn_marie, the researcher released two new SLQ quantizations of the Qwen3.8 27B model. One targets EAR 98, and the other aims to beat the official FP8 version while being about 10GB smaller. The post says GPTQ is near its limit at very low precision, and that more expensive methods are planned for Qwen4. These are the author's stated aims, and VQV has not independently tested the models.
AI-assisted summary based on the listed source.
Benjamin Marie released two new Qwen3.8 27B SLQ quantizations, one targeting EAR 98 and one aiming to beat the official FP8 while being about 10GB smaller. He says GPTQ is near its limit at very low precision and plans more expensive methods for Qwen4. Original post on X Section: Models & Releases
Developers who run open models on limited hardware could benefit if the smaller quantized versions match the official FP8 release. The post's note that GPTQ is near its limit suggests that further gains at very low precision may depend on costlier techniques. Watch for independent benchmark results and the planned Qwen4 work.
VQV organizes public signals from inspectable sources. It does not independently verify the underlying report.
Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.
VQV surfaced this signal because it is recent, relevant to Open Source LLMs, connected to Scobleizer News Pulse (X).
No login, cookies, social SDKs, or automatic posting.
