Kimi-VL-A3B-Thinking-2506
moonshotai's Kimi-VL-A3B-Thinking-2506 packs 3 billion parameters into a image-generation model. Kimi-VL-A3B-Thinking-2506 is an open-weights image model with roughly 3 billion parameters.
by moonshotai · 3B parameters
Best for
Ways to use Kimi-VL-A3B-Thinking-2506 in osFoundry
Connect with your own key (BYOK)
Open the key dialog and paste your moonshotai API key. osFoundry discovers Kimi-VL-A3B-Thinking-2506 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-VL-A3B-Thinking-2506 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-VL-A3B-Thinking-2506
Kimi-VL-A3B-Thinking-2506 runs on a single 16GB consumer GPU (~2 GB VRAM with KV-cache headroom). Full-precision inference fits on a single H100 80GB at FP16 precision (~8 GB).
Kimi-VL-A3B-Thinking-2506 vs similar models
Licence
Unspecified — Licence terms not specified — verify the upstream model card before commercial use.
Check upstream documentation.
Frequently asked about Kimi-VL-A3B-Thinking-2506
Is Kimi-VL-A3B-Thinking-2506 free to use?
Kimi-VL-A3B-Thinking-2506 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-VL-A3B-Thinking-2506 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-VL-A3B-Thinking-2506 need?
Approximately 2 GB at Q4 quantisation, or 8 GB at full FP16 precision. Fits on a single 24GB consumer GPU.
Can I run Kimi-VL-A3B-Thinking-2506 locally?
Yes. Kimi-VL-A3B-Thinking-2506 is open-weights and runs locally on a workstation GPU. osFoundry's local runtime handles model loading, quantisation, and routing.
What is Kimi-VL-A3B-Thinking-2506 best at?
Kimi-VL-A3B-Thinking-2506 is well-suited to image text to text.
How do I use Kimi-VL-A3B-Thinking-2506 in osFoundry?
Paste your moonshotai API key in the key dialog (or deploy the open weights for self-hostable models), assign Kimi-VL-A3B-Thinking-2506 to a Maestro role in the Pipeline tab, then use it in chat, Room Apps via invokeAI, or your own apps.
Published by moonshotai on June 21, 2025. Source: https://huggingface.co/moonshotai/Kimi-VL-A3B-Thinking-2506