Skip to content
localmodel.run

Text model · Flint

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

Flint 40B-A2.5B needs about 25.9 GB to run at Q4_K_M (the Q4_K_M GGUF file is ~23.61 GB to download; KV cache and overhead add the rest), or about 41.7 GB at Q8_0. The lightest hardware that runs it is Nvidia GeForce RTX 5090 (32GB).

Flint family · 39.8B params (Mixture-of-Experts: activates only 2.5B of 39.8B params per token, so generation is faster than the total size suggests) · released Oct 2026.

Shopping for hardware? See what runs Flint 40B-A2.5B →

Set up Flint 40B-A2.5B on your hardware →

License Apache-2.0 · Commercial OK ↓ 26.7K/mo ♥ 1.3K on HuggingFace
Q4_K_M GGUF
23.61 GB
Q8_0 GGUF
39.44 GB
Memory @ Q4 (4k)
~25.9 GB
Context
256 k

Q4_K_M and Q8_0 are GGUF download (file) sizes. Memory @ Q4 includes the KV cache and overhead, so it runs larger than the file on disk.

Will it run on your device?

Flint 40B-A2.5B runs on 10 of 49 tracked devices at Q4_K_M.

9 run well 1 tight fit 39 too small
Fit check Q4_K_M
No, not enough memory
needs 25.9 GB usable 10.5 GB
See the full breakdown

Memory breakdown

Weights (Q4_K_M)23.61 GB
+
KV cache (4k)1.5 GB
+
Overhead0.8 GB
=
Total25.9 GB

How context length changes it

4k context ~25.9 GB 32k context ~36.5 GB 128k context ~72.9 GB

Longer context grows the KV cache, which for Flint 40B-A2.5B is sized by its 2.5B active params, not the full 39.8B. It needs ~25.9 GB at 4k and ~72.9 GB at 128k.

Quantization sizes

GGUF quantson disk
Quantization Size on disk
Q2_K 16.7 GB est
Q3_K_M 19.5 GB est
Q4_K_M (default) 23.61 GB
Q5_K_M 28.4 GB est
Q6_K 32.6 GB est
Q8_0 39.44 GB
FP16 79.6 GB est

Lower quant = smaller and faster, slightly lower quality. Q4_K_M is the common default.

Run it

Ollama
$ ollama run flint:40b
llama.cpp
$ llama-cli -hf bartowski/RecursiveSI_flint-GGUF:Q4_K_M
LM Studio
$ lms get bartowski/RecursiveSI_flint-GGUF

Which devices can run Flint 40B-A2.5B?

NVIDIA GPUs

Apple Silicon Macs

RAM-only laptops

iPhone & iPad

Android

AMD GPUs

Same job, different size

Pick by what fits your memory: step down to free up VRAM, or step up if you have headroom.

Next size down Qwen3.6 35B-A3B 36B Next size up Mixtral 8x7B 46.7B

Similar models

FAQ

How much VRAM or RAM does Flint 40B-A2.5B need?

At Q4_K_M, Flint 40B-A2.5B needs about 25.9 GB (weights ~23.61 GB + KV cache + overhead) at a 4k context. At Q8_0 budget ~41.7 GB.

What is the Q4_K_M GGUF file size of Flint 40B-A2.5B?

The Q4_K_M GGUF file is about 23.61 GB to download, and the Q8_0 GGUF is about 39.44 GB. That is the weights file on disk; to run it you also need room for the KV cache and overhead, so budget ~25.9 GB of memory at Q4_K_M.

Can Flint 40B-A2.5B run on a laptop?

Flint 40B-A2.5B is large; you need a high-memory Mac or multi-GPU setup at Q4_K_M.

Is Flint 40B-A2.5B cheaper to run because it is a MoE model?

It is faster, not lighter. Flint 40B-A2.5B activates only 2.5B of 39.8B params per token (so it runs quickly), but all experts must stay in memory, so it still needs memory for the full 39.8B.

Can I use Flint 40B-A2.5B commercially?

Yes. Flint 40B-A2.5B is licensed Apache-2.0, which permits commercial use.

Understand the numbers

Short guides to the ideas behind Flint 40B-A2.5B's memory and quant figures.

Recursive coding MoE: 39.8B total and approximately 2.5B active including the output head. Q4_K_M (25,346,651,360 bytes) and Q8_0 (42,350,359,776 bytes) are measured bartowski GGUFs, converted to GiB. The separately measured Ollama default model is NVFP4 (22,863,084,160 bytes), not this Q4_K_M artifact. The vendor documents 262,144 tokens.

Sources

Catalog updated 2026-11-01. Memory figures are estimates. See methodology.