# Flint 40B-A2.5B: RAM and VRAM requirements

> Flint 40B-A2.5B is a 39.8B Flint model (Mixture-of-Experts, 2.5B active per token). At Q4_K_M it needs about **25.9 GB** to run and fits **10 of 49** tracked devices. Minimum to run: Nvidia GeForce RTX 5090 (32GB).

Catalog updated: 2026-11-01. Sources: Ollama, HuggingFace GGUF repos, vendor specs.

## Memory by quantization
| Quant | On disk | To run (4k context) |
| --- | --- | --- |
| Q4_K_M | 23.61 GB | ~25.9 GB |
| Q8_0 | 39.44 GB | ~41.7 GB |

Memory = weights + KV cache + ~0.8 GB runtime overhead, and varies ±15% with context length.

## Will it run on my device?
- **Nvidia GeForce RTX 2060 (6GB)** (6 GB): No, not enough memory
- **iPhone 17** (8 GB): No, not enough memory
- **iPhone 17 Pro** (12 GB): No, not enough memory
- **16GB RAM Laptop (CPU/iGPU only)** (16 GB): No, not enough memory
- **Samsung Galaxy S26 Ultra (16GB, 1TB config)** (16 GB): No, not enough memory
- **Apple M4 Pro (24GB)** (24 GB): No, not enough memory
- **Apple M4 Pro (48GB)** (48 GB): Yes, it runs
- **Apple M3 Ultra (256GB)** (256 GB): Yes, it runs (room for FP16)

Full table of all 49 devices: https://localmodel.run/model/flint-40b-a2.5b

## How to run
Quickest path: `ollama run flint:40b`. On Mac, LM Studio (ships MLX) is fastest; on Linux, Ollama for chat or vLLM to serve; on Windows, LM Studio or Ollama.

## Details
- Parameters: 39.8B (MoE, 2.5B active per token)
- Default context: 256k tokens
- License: Apache-2.0 (commercial use: yes)
- Released: 2026-10
- HuggingFace: 26,703 downloads/mo, 1285 likes

## FAQ
### How much VRAM or RAM does Flint 40B-A2.5B need?
About 25.9 GB at Q4_K_M (weights 23.61 GB + KV cache + overhead) at a 4k context. Budget ~41.7 GB for Q8_0.
### Can Flint 40B-A2.5B run on a laptop?
Flint 40B-A2.5B is large; you need a high-memory Mac or a 24 GB+ GPU at Q4_K_M.
### Can I use Flint 40B-A2.5B commercially?
Yes, Apache-2.0 permits commercial use.

Sources: https://huggingface.co/RecursiveSI/flint, https://huggingface.co/RecursiveSI/flint/blob/main/config.json, https://huggingface.co/bartowski/RecursiveSI_flint-GGUF, https://ollama.com/library/flint:40b
More: https://localmodel.run/model/flint-40b-a2.5b