Expired80% confidenceFactTime unknown
The Q5 version of OmniMerge V4 can support a context length of 100K.
1
Sources
80%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026(expired)
Sources
Related Entities
Related Claims
UnverifiedThe Q5 version of OmniMerge V4 runs noticeably faster on the RTX 5090 than the standard Q4 quantized version.78% similarUnverifiedOllama默认提供的模型多为4-bit量化版本（Q4_0或Q4_K_M格式）67% similarUnverifiedOmniMerge V4 is a model formed by fusing three to four 27B models through weight merging.66% similarUnverifiedQLoRA combines 4-bit quantization, making it possible to fine-tune 70B parameter models on consumer-grade GPUs.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/18012API
curl https://kongchang.com/api/v1/knowledge/claims/18012MCP
get_claim(id=18012)