Unverified50% confidenceBenchmarkExact time
On a MacBook Pro M5 Pro with 48 GB of memory and a 32K context window, the average time per search question in Markdown (restrictor) mode: Qwen 3.8 27B about 42 seconds, Meta Muse Glimmer 30B about 62 seconds, Prism Bonsai 27B about 24 seconds, and Gemma 4 12B QAT about 16 seconds (about 43 seconds after switching to Markdown Raw).
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/20/2026
First Seen
Valid until: 11/3/2026
Sources
Related Entities
Related Claims
UnverifiedRunning a 27B model on M1 Max with a 32,000-token prompt requires waiting more than 500 seconds before the first character is output, whereas the M5 Max takes less than one minute (approximately 40 seconds with optimized software).78% similarUnverifiedIn the time-to-first-token test on a 70-billion-parameter model, the M3 Ultra took 31.6 seconds while the M5 Ultra took only 7.7 seconds, achieving approximately 4.1x, close to Apple's claimed 4.3x.73% similarUnverifiedFor a cold-start 27k-token fresh session, the original llama.cpp required 245 seconds, the author's fork took 228 seconds on Mac alone, and was reduced to 168 seconds after connecting the phone.73% similarUnverifiedWhen generating images using a 4-billion-parameter distilled model, the Mac mini produces approximately one image every 12 seconds (about 5 per minute), while the Mac Studio produces approximately one image every 2 seconds (about 29 per minute).73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1074216API
curl https://kongchang.com/api/v1/knowledge/claims/1074216MCP
get_claim(id=1074216)