Unverified50% confidenceReleaseExact time
Qwen3.8 27B officially provides a BFloat16 full-precision version and an FP8 version; Unsloth has released a 4-bit version with NVFP4 quantization.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/16/2026
First Seen
Valid until: 12/15/2026
Sources
Related Entities
Related Claims
UnverifiedThe Qwen3.8-27B model is available in multiple formats including Safetensors, GGUF, NVFP4, NVFP4 GGUF variants, and GPTQ-Int4 quantized versions.78% similarUnverifiedAfter being quantized to 4-bit precision via tools such as Ollama, the 14B parameter version of Qwen3 can run smoothly on consumer-grade GPUs.72% similarUnverifiedThe Qwen 3.5-32B model is available on the Bailian platform.72% similarUnverifiedQwythos-9B is built on top of the uncensored Qwen 3.5 base model.69% similarUnverifiedOpen-source models such as LLaMA 3, Mistral, and Qwen 2.5 can already run smoothly on mainstream consumer-grade GPUs (such as the RTX 4090) at parameter scales ranging from 7B to 14B.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/924470API
curl https://kongchang.com/api/v1/knowledge/claims/924470MCP
get_claim(id=924470)