FreeToken

All projects

Apps & Tools

Edge-native MoE serving engine for running massive LLMs on low VRAM GPUs

by47thtechcorner

PlayableOpen source

FreeToken media is blocked

Allow external media to connect to the provider and play this content.

Description

FreeToken is an inference engine designed to run large frontier models like Qwen 3.8 27B on consumer hardware with as little as 4GB VRAM. It features a bandwidth-adaptive hybrid backend, dynamic VRAM re-allocation, and provides a drop-in OpenAI-compatible API endpoint for local agent workflows.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Details

Why this is interesting: It achieves a significant technical feat by serving a 27B parameter MoE model on consumer hardware with extremely…
Attributed to: @47thtechcorner
Built with: FLUX 3 Video K2.7 Code North Mini Code Grok Video 1.5 Qwen3.8-27B Gemini 3.7 Flash Grok 4.6 DeepSeek-V4-Flash-0731 Qwen3.7 Flash Opus 5 3.6 Flash 3.5 Flash Cyber Kimi K3 GLM-5.2 Step 3.7 Flash Opus 4.8 Gemini 3.5 Flash V4 Opus 4.7 Qwen3.6 Opus 4.6 Opus 4.5 2.5 Flash Qwen3 30B-A3B Qwen3 8B Qwen3 14B
Links: flashml.ai
View on YouTube
View GitHub repo
Tags: #llm #moe #local-ai #inference

Discovered: X
Listed: Added 3d ago
Build: Strong
Models: Strong

Prompt

Procedural voxel islands with waterfalls, apple trees and wandering fireflies - rendered in real time. No assets, just code.

Source

FreeToken

Description

Timeline

FreeToken

Description

Timeline