K-intelligenceLlamaForCausalLM

Midm 2.0 Base Instruct — Hardware Requirements & GPU Compatibility

Chat

Midm 2.0 Base Instruct is a 11.5B-parameter open language model from K-intelligence. It supports a context window of up to 32,768 tokens. At Q4_K_M it needs about 7.63 GB of VRAM — see which GPUs and Macs can run it below.

11.5B params · dense33K contextMITJul 202583.3K downloads141 likes387 quant downloads

Your hardware

Verdicts below are for the reference RTX 4090. Set yours to personalise the page.

Detection reads your browser's GPU name with no data sent anywhere.

Runs great on a 24 GB GPU

Runs from 8 GB (Q4_K_M · 2K)Best on 24 GB (Q4_K_M · ~86 tok/s · 32K)Macs from 32 GB (~11 tok/s)

MIT

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Midm 2.0 Base Instruct

One recommended way. Every other method, OS and quant is in the step-by-step guide.

Other ways

each opens step-by-step

Run in cloud

Fits on Tesla V100 PCIe 16GB (5 GB headroom) · Q4_K_M

Generation speed: ~77 tok/s; generation speed
Cost per 1M output tokens: $0.29; per 1M output tokens

Deployon Vast.aiBest optionfrom $0.080/hr

Compare GPUs →

Run Midm 2.0 Base Instruct

10.5 GB VRAM at 16K context · fits the RTX 4090 (13.5 GB free)

Works on macOS, Windows and Linux.

Install Ollama (Linux)
$curl -fsSL https://ollama.com/install.sh | sh
Install Ollama (macOS / Windows)
Download and run the app; it installs the ollama command.
Download Ollama ↗
Run Q4_K_M
$ollama run hf.co/mykor/Midm-2.0-Base-Instruct-gguf:Q4_K_M
Raise the context window
Type it inside the chat. For the API server instead, start it with OLLAMA_CONTEXT_LENGTH=16384 ollama serve.
$/set parameter num_ctx 16384
A larger context needs more memory than the weights alone.
Use the API
Ollama serves its API on http://localhost:11434.

Works on macOS, Windows and Linux.

Install LM Studio
Desktop app for macOS (Apple silicon), Windows and Linux.
lmstudio.ai ↗
Download the model
In the app press Ctrl/Cmd + Shift + M and paste https://huggingface.co/mykor/Midm-2.0-Base-Instruct-gguf, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get https://huggingface.co/mykor/Midm-2.0-Base-Instruct-gguf@Q4_K_M
Open on Hugging Face ↗
Load and chat
Open the model loader, select the model and set Context Length to 16384. Leave GPU offload at maximum if it fits.
Start the local server (optional)
$lms server start

Works on macOS, Windows and Linux.

Install llama.cpp (macOS)
MacPorts, Nix and conda-forge also work.
$brew install llama.cpp
Install options ↗
Install llama.cpp (Windows)
$winget install llama.cpp
Install llama.cpp (Linux)
Download the prebuilt llama-*-bin-ubuntu-* archive for your GPU backend (CPU, Vulkan) from the releases page, unpack it and run the binaries from that folder. Homebrew (brew install llama.cpp) and conda-forge also work on Linux.
llama.cpp releases ↗
Start the server
Downloads the model on first run. The chat UI and an OpenAI-compatible API are at http://localhost:8080.
$llama-server -hf mykor/Midm-2.0-Base-Instruct-gguf:Q4_K_M -c 16384 -ngl 99 --port 8080

Works on macOS, Windows and Linux.

Option A: Docker Model Runner
Docker Desktop: Settings > AI > Enable Docker Model Runner. Docker Engine on Linux: install the docker-model-plugin package.
$docker model run hf.co/mykor/Midm-2.0-Base-Instruct-gguf:Q4_K_M
Model Runner docs ↗
The quant tag after the repo is how Hugging Face cards write it; if your Docker version rejects it, drop the tag and pick the quant by repo.
Option B: llama.cpp server image (NVIDIA)
Needs the NVIDIA Container Toolkit. Models are cached in the llama-cache volume. Open http://localhost:8080.
$docker run --gpus all -p 8080:8080 -v llama-cache:/root/.cache ghcr.io/ggml-org/llama.cpp:server-cuda -hf mykor/Midm-2.0-Base-Instruct-gguf:Q4_K_M -c 16384 -ngl 99 --host 0.0.0.0 --port 8080
Image tags (cuda13, vulkan, rocm) ↗
Cache path inside the image is not in the docs; remove the -v flag if the volume complains.

Linux only.

Install vLLM
Linux only; on Windows run it inside WSL2. NVIDIA GPU shown; AMD ROCm, Intel and CPU builds have their own install pages.
$uv venv --python 3.12 --seed
$source .venv/bin/activate
$uv pip install vllm --torch-backend=auto
vLLM installation ↗
Serve the model
OpenAI-compatible API at http://localhost:8000/v1.
$vllm serve K-intelligence/Midm-2.0-Base-Instruct --max-model-len 16384
Runs the original ~25 GB weights, not the Q4_K_M GGUF.

macOS only.

MLX supports only macOS, so follow these steps for that setup.

Install mlx-lm
$pip install mlx-lm
Chat in the terminal
$mlx_lm.chat --model litmudoc/Midm-2.0-Base-Instruct-MLX-Q8
Downloads the model on first run. The MLX repo has its own quantisation; it may not match the quant selected above.
Or start a local server
Listens on http://localhost:8080.
$mlx_lm.server --model litmudoc/Midm-2.0-Base-Instruct-MLX-Q8

Works on macOS, Windows and Linux.

Install Jan
jan.ai ↗
Find the model
Open the Hub tab, paste this model ID in Search, and choose Q4_K_M if offered.
$mykor/Midm-2.0-Base-Instruct-gguf
Use the local API (optional)
Settings > Local API Server > Start Server. It listens on http://127.0.0.1:1337.

Rent GPUs hourly to run Q4_K_M right there. Pricing comes from partner marketplaces and varies.

Available GPU rentals for Q4_K_M, prioritized by value
GPU	Speed	$/1M tokens	Rent
Tesla V100 PCIe 16GBCheapestCheapest5 GB free	~77 tok/s	$0.29	Deploy on Vast.aifrom $0.080/hr (opens in new tab)	5 GB free · ~77 tok/s · $0.29 / 1M tokens
RTX A4000 (16 GB)5 GB free	~38 tok/s	$0.76	Deploy on Vast.aifrom $0.10/hr (opens in new tab)	16 GB · 5 GB free · ~38 tok/s · $0.76 / 1M tokens
RTX 4060 Ti 16GB5 GB free	~25 tok/s	$1.84	Deploy on Vast.aifrom $0.16/hr (opens in new tab)	5 GB free · ~25 tok/s · $1.84 / 1M tokens
RTX 5060 Ti 16GB5 GB free	~38 tok/s	$1.36	Deploy on Vast.aifrom $0.19/hr (opens in new tab)	5 GB free · ~38 tok/s · $1.36 / 1M tokens
RTX A6000 (48 GB)37 GB free	~65 tok/s	$1.40	Deploy on RunPodfrom $0.33/hr (opens in new tab)	48 GB · 37 GB free · ~65 tok/s · $1.40 / 1M tokens
RTX PRO 4500 (32 GB)21 GB free	~76 tok/s	$1.24	Deploy on RunPodfrom $0.34/hr (opens in new tab)	32 GB · 21 GB free · ~76 tok/s · $1.24 / 1M tokens
RTX 4090 (24 GB)13 GB free	~86 tok/s	$1.10	Deploy on RunPodfrom $0.34/hr (opens in new tab)	24 GB · 13 GB free · ~86 tok/s · $1.10 / 1M tokens
A40 (48 GB)37 GB free	~59 tok/s	$1.64	Deploy on RunPodfrom $0.35/hr (opens in new tab)	48 GB · 37 GB free · ~59 tok/s · $1.64 / 1M tokens

Compare GPUs →

How Much VRAM Does Midm 2.0 Base Instruct Need?

Select a quantization level; the listed hardware adheres to it.

VRAM	+32K	+128K	File	Quality	On RTX 4090 (reference)
5.6 GB	11.7 GB	—	4.5 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Fits: 32K context · ~117 tok/s
5.8 GB	11.8 GB	—	5.2 GB	3-bit small — 3-bit small quantization	Fits: 32K context · ~114 tok/s
6.3 GB	12.4 GB	—	5.7 GB	3-bit medium — 3-bit medium quantization	Fits: 32K context · ~104 tok/s
6.5 GB	12.5 GB	—	6.6 GB	4-bit · legacy — 4-bit legacy quantization	Fits: 32K context · ~101 tok/s
6.6 GB	12.7 GB	—	6.2 GB	3-bit large — 3-bit large quantization	Fits: 32K context · ~99 tok/s
6.9 GB	13 GB	—	6.4 GB	4-bit · i-quant compact — Importance-weighted 4-bit, compact	Fits: 32K context · ~95 tok/s
7.2 GB	13.2 GB	—	6.7 GB	4-bit small — 4-bit small quantization	Fits: 32K context · ~91 tok/s
7.2 GB	13.2 GB	—	6.7 GB	4-bit · i-quant non-linear — Importance-weighted 4-bit, non-linear	Fits: 32K context · ~91 tok/s
7.6 GB	13.7 GB	—	7 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Fits: 32K context · ~86 tok/s
7.9 GB	14 GB	—	8 GB	5-bit · legacy — 5-bit legacy quantization	Fits: 32K context · ~83 tok/s
8.6 GB	14.7 GB	—	8 GB	5-bit small — 5-bit small quantization	Fits: 32K context · ~76 tok/s
8.9 GB	15 GB	—	8.2 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Fits: 32K context · ~73 tok/s
10.2 GB	16.3 GB	—	9.5 GB	6-bit, very good — 6-bit quantization, very good quality	Fits: 32K context · ~64 tok/s
12.3 GB	18.3 GB	—	12.3 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Fits: 32K context · ~54 tok/s
23.8 GB	29.8 GB	—	23.1 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Fits: 2K context · ~28 tok/s

Selected Q4_K_M

4-bit medium quantization

10.5GBat 16K context

Weights: 6.9 GB
KV cache: 3.2 GB
Overhead: 0.3 GB

Fits RTX 4090 (reference) · 13.6 GB free · ~86 tok/s

≥50 tok/s needs ≥586 GB/s · RTX 5090 ~153, RTX 3090 Ti ~86

Which Hardware Can Run Midm 2.0 Base Instruct?

Q4_K_M · 7.6 GB

Midm 2.0 Base Instruct (Q4_K_M) requires 7.6 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 10+ GB is recommended. Using the full 33K context window can add up to 6.0 GB, bringing total usage to 13.7 GB. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti, NVIDIA GeForce RTX 3080. 156 devices with unified memory can run Midm 2.0 Base Instruct, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, iPad Pro M5 13" (16 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M

Typical hardwareQuantVRAMtok/sContext that fits

8 GB · RTX 4060Q4_K_M · 10.5 GB · ~23 tok/s · 2Ktight10.5~232K8 GB tier →

12 GB · RTX 3060 12GBQ4_K_M · 10.5 GB · ~31 tok/s · 16K10.5~3116K12 GB tier →

16 GB · RTX 5060 Ti 16GBQ4_K_M · 10.5 GB · ~38 tok/s · 32K10.5~3832K16 GB tier →

24 GB · RTX 4090referenceQ4_K_M · 10.5 GB · ~86 tok/s · 32K10.5~8632K24 GB tier →

Mac 32 GB · Mac Mini M4Q4_K_M · 10.5 GB · ~11 tok/s · 32K10.5~1132KMacs →

Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 10.5 GB · ~50 tok/s · 32K10.5~5032KMacs →

Alternatives to Midm 2.0 Base Instruct

Similar models

6tok/s on RTX 4090

Model	Score	VRAM	tok/s	Why it is similar
Gemma 3 12B ITGoogle	74.2	10.9 GB	~82	same size
Gemma 4 12BGoogle	—	9 GB	~80	same size
Qwen2.5 Omni 7BAlibaba	—	9.9 GB	~93	same size
Mistral Nemo Base 2407Mistral AI	—	11 GB	~81	same size
Falcon 11BTII UAE	—	8.5 GB	~89	same size
NVIDIA Nemotron Nano 12B v2NVIDIA	—	11 GB	~81	same size

Specifications

Publisher: K-intelligence
Parameters: 11.5B
Architecture: LlamaForCausalLM
Context Length: 32,768 tokens
Vocabulary Size: 131,384
Release Date: 2025-07-03
License: MIT

Resources

Hugging Face model card (opens in a new tab)

Frequently Asked Questions

How much VRAM does Midm 2.0 Base Instruct need?

Midm 2.0 Base Instruct requires 7.6 GB of VRAM at Q4_K_M, or 23.8 GB at BF16. Full 33K context adds up to 6.0 GB (13.7 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 11.5B × 4.8 bits ÷ 8 = 6.9 GB

KV Cache + Overhead ≈ 0.7 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 6.8 GB (at full 33K context)

VRAM usage by quantization

Q4_K_M

7.6 GB

Q4_K_M + full context

13.7 GB

Learn more about VRAM estimation →

What's the best quantization for Midm 2.0 Base Instruct?

For Midm 2.0 Base Instruct, Q4_K_M (7.6 GB) offers the best balance of quality and VRAM usage. Q5_0 (7.9 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 5.6 GB.

VRAM requirement by quantization

Q2_K

5.6 GB~75%

Q3_K_L

6.6 GB~86%

IQ4_NL

7.2 GB~88%

Q4_K_M ★

7.6 GB~89%

Q5_K_M

8.9 GB~92%

BF16

23.8 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Midm 2.0 Base Instruct on a Mac?

Yes, but only at lower quantizations. The smallest Mac that can run Midm 2.0 Base Instruct is Apple iPhone 17 Pro (12 GB) at Q2_K; 139 of the 156 Macs we list can run it at some quantization. For Q4_K_M (7.6 GB) you need a Mac with more unified memory.

Can I run Midm 2.0 Base Instruct locally?

Yes — Midm 2.0 Base Instruct can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Midm 2.0 Base Instruct?

At Q4_K_M, Midm 2.0 Base Instruct can reach ~629 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~86 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = (bandwidth GB/s ÷ model GB) × efficiency

Example: NVIDIA B200 → 8000 ÷ 7.63 × 0.65 = ~682 tok/s

Estimated speed at Q4_K_M (7.6 GB)

NVIDIA B200

~682 tok/s

NVIDIA GeForce RTX 4090

~86 tok/s

NVIDIA B300

~682 tok/s

AMD Instinct MI350X

~629 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of Midm 2.0 Base Instruct?

At Q4_K_M, the download is about 6.93 GB. The full-precision BF16 version is 23.09 GB. The smallest option (Q2_K) is 4.91 GB.

Which GPUs can run Midm 2.0 Base Instruct?

52 consumer GPUs can run Midm 2.0 Base Instruct at Q4_K_M. Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT, AMD Radeon RX 7600. 36 GPUs have plenty of headroom for comfortable inference.

Which devices can run Midm 2.0 Base Instruct?

156 devices with unified memory can run Midm 2.0 Base Instruct at Q4_K_M (7.6 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Asus ROG Flow Z13 (2025, Ryzen AI Max+ 395, 128 GB), Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.