Kimi-Linear-48B-A3B-Instruct-mlx-bf16
mlx-community's Kimi-Linear-48B-A3B-Instruct-mlx-bf16 packs 48 billion parameters into a chat model. Kimi-Linear-48B-A3B-Instruct-mlx-bf16 is an open-weights chat model with roughly 48 billion parameters.
by mlx-community · 48B parameters
Best for
Ways to use Kimi-Linear-48B-A3B-Instruct-mlx-bf16 in osFoundry
Connect with your own key (BYOK)
Open the key dialog and paste your mlx-community API key. osFoundry discovers Kimi-Linear-48B-A3B-Instruct-mlx-bf16 automatically — assign it to a Maestro role (router, direct, orchestrator, or fallback) in the Pipeline tab and it is live in every chat. Your key, your provider account — no token markup.
Deploy a dedicated endpoint
Kimi-Linear-48B-A3B-Instruct-mlx-bf16 is open-weights — run it locally for free, or deploy a dedicated GPU endpoint in your workspace for reserved capacity with no rate limits.
Use it in a Room App
Room Apps declare AI features in their manifest, then call them with invokeAI:
import { invokeAI } from '@osfoundry/app-sdk'
// 'summarize' is an AI feature declared in your app manifest.
const result = await invokeAI('summarize', userText)
Call it from your own apps
Once a model is wired into your workspace you can host it as an API and reach it from your own services, scripts, or CI — outside osFoundry.
What hardware can run Kimi-Linear-48B-A3B-Instruct-mlx-bf16
Kimi-Linear-48B-A3B-Instruct-mlx-bf16 runs on a single A100 40GB at Q4 quantisation (~29 GB VRAM with KV-cache headroom). Full-precision inference requires an H200 141GB or 2x A100 80GB at FP16 (~116 GB).
Kimi-Linear-48B-A3B-Instruct-mlx-bf16 vs similar models
Licence
Unspecified — Licence terms not specified — verify the upstream model card before commercial use.
Check upstream documentation.
Frequently asked about Kimi-Linear-48B-A3B-Instruct-mlx-bf16
Is Kimi-Linear-48B-A3B-Instruct-mlx-bf16 free to use?
Kimi-Linear-48B-A3B-Instruct-mlx-bf16 is free to run locally on your own hardware. Hosted access through osFoundry is metered (input Free (local), output Free (local)). You can switch between local and hosted at any time.
Can I use Kimi-Linear-48B-A3B-Instruct-mlx-bf16 commercially?
Commercial use is allowed with conditions. Licence terms not specified — verify the upstream model card before commercial use. Check upstream documentation.
How much VRAM does Kimi-Linear-48B-A3B-Instruct-mlx-bf16 need?
Approximately 29 GB at Q4 quantisation, or 116 GB at full FP16 precision. Fits on a single A100/H100 80GB.
Can I run Kimi-Linear-48B-A3B-Instruct-mlx-bf16 locally?
Yes. Kimi-Linear-48B-A3B-Instruct-mlx-bf16 is open-weights and runs locally on a workstation GPU. osFoundry's local runtime handles model loading, quantisation, and routing.
What is Kimi-Linear-48B-A3B-Instruct-mlx-bf16 best at?
Kimi-Linear-48B-A3B-Instruct-mlx-bf16 is well-suited to text generation.
How do I use Kimi-Linear-48B-A3B-Instruct-mlx-bf16 in osFoundry?
Paste your mlx-community API key in the key dialog (or deploy the open weights for self-hostable models), assign Kimi-Linear-48B-A3B-Instruct-mlx-bf16 to a Maestro role in the Pipeline tab, then use it in chat, Room Apps via invokeAI, or your own apps.
Published by mlx-community on November 24, 2025. Source: https://huggingface.co/mlx-community/Kimi-Linear-48B-A3B-Instruct-mlx-bf16