LLM EXPLORER 65,548 MODELS INDEXED

Flint MLX Serve Mixed 4 8bit by pgmharikrishnan

By pgmharikrishnan · 1230 downloads
HF Repo ↗ Copy URL

Flint MLX Serve Mixed 4 8bit is an open-source language model by pgmharikrishnan. Features: 39.8b LLM, VRAM: 23.4GB, Context: 256K, License: apache-2.0, Quantized, LLM Explorer Score: 0.35.

  8-bit   8bit   Apple-silicon Base model:quantized:recursive...   Base model:recursivesi/flint   Conversational   Flint   Mixed-4-8bit   Mlx   Mlx-serve   Moe   Quantized   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/pgmharikrishnan/Flint-MLX-Serve-mixed-4-8bit 

Flint MLX Serve Mixed 4 8bit Parameters and Internals

LLM NameFlint MLX Serve Mixed 4 8bit
Repository 🤗https://huggingface.co/pgmharikrishnan/Flint-MLX-Serve-mixed-4-8bit 
Base Model(s)  Flint   RecursiveSI/flint
Model Size39.8b
Required VRAM23.4 GB
Updated2026-10-25
Maintainerpgmharikrishnan
Model Typeflint
Model Files  4.4 GB: 1-of-6   4.4 GB: 2-of-6   4.4 GB: 3-of-6   4.3 GB: 4-of-6   4.3 GB: 5-of-6   1.6 GB: 6-of-6
Quantization Type8bit
Model ArchitectureFlintForCausalLM
Licenseapache-2.0
Context Length262144
Model Max Length262144
Transformers Version5.12.1
Tokenizer ClassPreTrainedTokenizerFast
Vocabulary Size150144
Torch Data Typebfloat16

Best Alternatives to Flint MLX Serve Mixed 4 8bit

Best Alternatives
Context / RAM
Downloads
Likes
Flint MLX Serve 4bit256K / 22.5 GB19502
Flint256K / 80 GB4795754
Flint FP8 BF16256K / 0.8 GB461
Note: green Score (e.g. "73.2") means that the model is better than pgmharikrishnan/Flint-MLX-Serve-mixed-4-8bit.