@pratikg says the 4-bit Inkling 2 Mini (17.6 GB) used 26 GB of memory at 128k context on a 64GB M5 Pro Mac mini, ran at about 30 tok/s at 512 tokens and 17 tok/s at 128k, was about...
VQV Signal
Update: Inkling 2 Mini gets hands-on local benchmarks and lands in LM Studio
@pratikg says the 4-bit Inkling 2 Mini (17.6 GB) used 26 GB of memory at 128k context on a 64GB M5 Pro Mac mini, ran at about 30 tok/s at 512 tokens and 17 tok/s at 128k, was about...
@pratikg says the 4-bit Inkling 2 Mini (17.6 GB) used 26 GB of memory at 128k context on a 64GB M5 Pro Mac mini, ran at about 30 tok/s at 512 tokens and 17 tok/s at 128k, was about 1.8x faster than the similar-size Qwen3.8-27B on short prompts, and passed all of the poster's small math, code and retrieval evals....
VQV organizes public signals from inspectable sources. It does not independently verify the underlying report.
Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.
Public Interest components
VQV surfaced this signal because it is recent, relevant to Open Source LLMs, connected to Scobleizer News Pulse (X).
No login, cookies, social SDKs, or automatic posting.