chatdb·LlamaForCausalLM

Natural Sql 7B — Hardware Requirements & GPU Compatibility

Chat

Natural Sql 7B is a 6.9B-parameter open language model from chatdb. It supports a context window of up to 4,096 tokens. At Q4_K_M it needs about 5.45 GB of VRAM — see which GPUs and Macs can run it below.

124.9K downloads 135 likes 1.6K quant downloads4K context

Specifications

Publisher
chatdb
Parameters
6.9B
Architecture
LlamaForCausalLM
Context Length
4,096 tokens
Vocabulary Size
102,400
Release Date
2024-02-02
License
CC BY-SA 4.0

Get Started

How Much VRAM Does Natural Sql 7B Need?

Select a quantization to see compatible GPUs below.

QuantizationBitsVRAM
Q2_K3.404.2 GB
Q3_K_S3.504.3 GB
Q3_K_M3.904.7 GB
Q4_04.004.8 GB
Q4_K_M4.805.5 GB
Q5_K_M5.706.2 GB
Q6_K6.607.0 GB
Q8_08.008.2 GB

Which GPUs Can Run Natural Sql 7B?

Q4_K_M · 5.5 GB

Natural Sql 7B (Q4_K_M) requires 5.5 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 8+ GB is recommended. Using the full 4K context window can add up to 1.0 GB, bringing total usage to 6.5 GB. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti, NVIDIA GeForce RTX 3070 Ti.

Runs great

— Plenty of headroom

Which Devices Can Run Natural Sql 7B?

Q4_K_M · 5.5 GB

58 devices with unified memory can run Natural Sql 7B, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, MacBook Air 13" M3 (8 GB).

Runs great

— Plenty of headroom
NVIDIA DGX H100~3196 tok/sNVIDIA DGX A100 640GB~1946 tok/sMac Studio (M3 Ultra, 256GB)~105 tok/sMac Studio (M3 Ultra, 512GB)~105 tok/sMac Studio (M3 Ultra, 96GB)~105 tok/sMac Pro M2 Ultra (192 GB)~103 tok/sMac Studio M2 Ultra (192 GB)~103 tok/sMacBook Pro 16" M5 Max (128 GB)~79 tok/sMac Studio M4 Max (128 GB)~70 tok/sMac Studio M4 Max (64 GB)~70 tok/sMacBook Pro 16" M4 Max (48 GB)~70 tok/sMacBook Pro 16" M4 Max (64 GB)~70 tok/sMac Studio M4 Max (36 GB)~53 tok/sMacBook Pro 14" M4 Max (36 GB)~53 tok/sMacBook Pro 16" M3 Max (48 GB)~53 tok/sMacBook Pro 14-inch (M5 Pro)~39 tok/sMac Mini M4 Pro (24 GB)~35 tok/sMac Mini M4 Pro (48 GB)~35 tok/sMacBook Pro 14" M4 Pro (24 GB)~35 tok/sMacBook Pro 16" M4 Pro (24 GB)~35 tok/sASUS Ascent GX10~33 tok/sNVIDIA DGX Spark~33 tok/sNVIDIA Jetson AGX Thor Developer Kit~33 tok/sAsus ROG Flow Z13 (2025, Ryzen AI Max+ 395, 128 GB)~31 tok/sBeelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB)~31 tok/sFramework Desktop (Ryzen AI Max+ 395, 128 GB)~31 tok/sGMKtec EVO-X2 (Ryzen AI Max+ 395, 128 GB)~31 tok/sHP Z2 Mini G1a (Ryzen AI Max+ PRO 395, 128 GB)~31 tok/sHP ZBook Ultra G1a 14 (Ryzen AI Max+ PRO 395, 128 GB)~31 tok/sMinisforum MS-S1 MAX (Ryzen AI Max+ 395, 128 GB)~31 tok/sSnapdragon X2 Elite Extreme Copilot+ PC~27 tok/sNVIDIA Jetson AGX Orin 32GB~24 tok/sNVIDIA Jetson AGX Orin 64GB~24 tok/sMacBook Pro 14-inch (M5)~20 tok/siPad Pro M5 13" (16 GB)~20 tok/sSnapdragon X Elite Copilot+ PC~16 tok/sMac Mini M4 (16 GB)~15 tok/sMac Mini M4 (32 GB)~15 tok/sMacBook Air 13" M4 (16 GB)~15 tok/sMacBook Air 13" M4 (24 GB)~15 tok/sMacBook Air 15" M4 (16 GB)~15 tok/sMacBook Air 15" M4 (24 GB)~15 tok/sMacBook Pro 14" M4 (16 GB)~15 tok/siPad Pro M4 13" (16 GB)~15 tok/sAMD Ryzen AI 9 HX 370 (Strix Point) Laptop~13 tok/sMacBook Air 13" M3 (16 GB)~13 tok/sMacBook Air 13" M3 (24 GB)~13 tok/sIntel Core Ultra 9 288V (Lunar Lake) Laptop~13 tok/sNVIDIA Jetson Orin NX 16GB~12 tok/s

Where to Download Natural Sql 7B

Community quantizations of this model — GGUF for llama.cpp, Ollama, and LM Studio, plus AWQ/MLX variants where available.

Related Models

Frequently Asked Questions

How much VRAM does Natural Sql 7B need?

Natural Sql 7B requires 5.5 GB of VRAM at Q4_K_M, or 15.1 GB at FP16. Full 4K context adds up to 1.0 GB (6.5 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 6.9B × 4.8 bits ÷ 8 = 4.1 GB

KV Cache + Overhead ≈ 1.4 GB (at 2K context + ~0.3 GB framework)

KV Cache + Overhead ≈ 2.4 GB (at full 4K context)

VRAM usage by quantization

5.5 GB
6.5 GB

Learn more about VRAM estimation →

What's the best quantization for Natural Sql 7B?

For Natural Sql 7B, Q4_K_M (5.5 GB) offers the best balance of quality and VRAM usage. Q5_K_S (6.1 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 3.2 GB.

VRAM requirement by quantization

IQ2_XXS
3.2 GB
IQ3_XS
4.2 GB
Q3_K_M
4.7 GB
Q4_K_M ★
5.5 GB
Q5_K_S
6.1 GB
FP16
15.1 GB

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Natural Sql 7B on a Mac?

Natural Sql 7B requires at least 3.2 GB at IQ2_XXS, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.

Can I run Natural Sql 7B locally?

Yes — Natural Sql 7B can run locally on consumer hardware. At Q4_K_M quantization it needs 5.5 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Natural Sql 7B?

At Q4_K_M, Natural Sql 7B can reach ~881 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~120 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 ÷ 5.5 × 0.65 = ~954 tok/s

Estimated speed at Q4_K_M (5.5 GB)

~954 tok/s
~120 tok/s
~954 tok/s
~881 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 Natural Sql 7B?

At Q4_K_M, the download is about 4.15 GB. The full-precision FP16 version is 13.82 GB. The smallest option (IQ2_XXS) is 1.90 GB.

Which GPUs can run Natural Sql 7B?

52 consumer GPUs can run Natural Sql 7B at Q4_K_M (5.5 GB). Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT, AMD Radeon RX 7600. 40 GPUs have plenty of headroom for comfortable inference.

Which devices can run Natural Sql 7B?

59 devices with unified memory can run Natural Sql 7B at Q4_K_M (5.5 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Apple iPhone 17 Pro, Asus ROG Flow Z13 (2025, 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.