Nicheloom

The opportunity tracker for new startups.

coachwhip

An 80B model, running comfortably on a 16 GB Mac. A mixture-of-experts inference engine for Apple silicon.

This is 1 of 97 launches in local and on-device AI runtimes — see how it stacks up on momentum and crowding →

1607 other launches read as similar to this one →

Details

External ID
1400083847
Source
GITHUB
Company
—
Product
coachwhip
Website domain
github.com
Launched
Oct. 1, 2026
Cohort
—
Upvotes
25
Upvotes percentile
0.5388711395101171
Tags
—
Fetched at
Oct. 5, 2026, 1:02 a.m.
Updated at
Oct. 5, 2026, 1:02 a.m.

Enrichment

Theme
local and on-device AI runtimes
Vertical
Horizontal
Function
Model & infra
Audience
Developer
AI stance
AI-native
Project type
Hobby / open-source project
Normalized one-liner
mixture-of-experts inference engine for apple silicon
Manually corrected
False

Could you build this?

No Executing an 80B MoE model on a 16 GB Apple Silicon Mac requires novel systems engineering, low-level Metal compute kernels, and specialized memory-streaming architectures that AI coding tools cannot design autonomously.

What it would actually take: This requires deep ML systems engineering in C++, Metal Shading Language (MSL), and Swift. The system needs custom quantized GEMM kernels, fast expert-weight streaming via asynchronous unified memory paging/SSD offloading, and aggressive activation caching tuned specifically for Apple Silicon's unified memory bandwidth.

Competitors

Other products that read as similar to this one — 1607 launches clear the similarity bar, closest 8 shown.

Attention rank: #748 of 1608 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).

Launched 337 days after the earliest competitor.

Other launches for this product

Same idea, different domain

Nobody's really built a model & infra tool for Fintech yet.