Unverified50% confidenceFactExact time
Pulling a model in OLLAMA without specifying a tag defaults to the Q4KM version (18 GB), with the tag being latest.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/16/2026
First Seen
Valid until: 12/15/2026
Sources
Related Entities
Related Claims
UnverifiedOllama uses quantization schemes such as Q4_K_M by default, reducing the VRAM requirement for a 31B-parameter model from approximately 62GB (FP16) to approximately 20GB.69% similarUnverifiedDeploying Qwen3 27B with 16 GB of VRAM limits the choice to the Q3_K_XL GGUF quantization version, which is the highest precision that 16 GB of VRAM can accommodate.66% similarUnverifiedWhen VRAM is less than 16GB, it is recommended to consider Q2 or switch to a smaller parameter model; 1-bit is not recommended.64% similarVerifiedThe Qwen2.5:7B model is approximately 4 GB in size after Q4 quantization.63% similarUnverified32GB of VRAM can run FP16 and FP8 original models directly, without needing to use GGUF quantized versions.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/925946API
curl https://kongchang.com/api/v1/knowledge/claims/925946MCP
get_claim(id=925946)