#llama-cpp

2 articles tagged

Chart-style graphic comparing LFM2.5-DSpark speedups on H100 and Apple M4 Max, August 2026

Engineering · Aug 22, 2026 · 12 min read

LFM2.5-DSpark ships 3.2x on an H100 and 1.18x on a MacBook: the 20 August 2026 numbers mobile teams should read

Liquid AI released LFM2.5-DSpark speculative-decoding drafters on 20 August 2026. The headline 3.2x is a single H100 cell. On an M4 Max the 2.6B mean is 2.27x and the 8B MoE mean is 1.18x.

Manu Shukla

#on-device AI #speculative decoding #LFM2.5

Chart-style hero comparing Apple Silicon memory bandwidth and measured tokens per second against hosted API token costs

AI & Cloud Infrastructure · Aug 4, 2026 · 14 min read

Apple Silicon vs cloud APIs in 2026: the local LLM break-even, with real benchmarks

Published benchmark data, Apple's own specifications and one hosted price list are enough to work out when buying a Mac beats renting tokens. The answer depends on prefill, not generation, and most of the M5 numbers

Manu Shukla

#local LLM #Apple Silicon #inference-cost