Unverified50% confidenceFactExact time
FollowUp Local uses the 4B model by default, and older laptops can switch to the 1B model via an environment variable.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/15/2026
First Seen
Valid until: 1/13/2027
Sources
Related Entities
Related Claims
UnverifiedEven when model weights are quantized to Int4, the KV Cache may still use FP16 or BF16, and in long-text high-concurrency scenarios the cache can become the largest consumer of GPU memory.60% similarUnverifiedFor beginner users, small models with 3B to 7B parameters (such as Qwen2.5-7B and Llama3.2-3B) offer the best cost-effectiveness as a starting point, and can run smoothly on an ordinary laptop with 8 GB of memory.57% similarVerifiedA 7B-parameter model requires only approximately 4 GB of VRAM to run after 4-bit quantization.57% similarUnverifiedOn consumer-grade hardware, a small GQA model with a large context window is generally preferable to a larger model locked to a 4K context window.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1033313API
curl https://kongchang.com/api/v1/knowledge/claims/1033313MCP
get_claim(id=1033313)