Back to RunThisModel

Best AI Models for GTX 1650

4GB VRAM • nvidia • turing

VRAM
4GB
Generation
turing
MSRP
$149
Vendor
nvidia
Runs Perfectly (S/A/B)
74
With Offloading (C/D)
25
Cannot Run (F)
10
S

Excellent (65 models)

Runs great with room for large context windows

Llama 3.2 3B Instruct

Meta

S
3.2BChat / LLMQ5_K_M2.7GB

Qwen 2.5 3B

Alibaba

S
3BChat / LLMQ4_K_M2.5GB

Qwen 2.5 Coder 3B

Alibaba

S
3BCodingQ4_K_M2.5GB

Falcon 3 3B

TII

S
3BChat / LLMQ4_K_M2.4GB

StableLM Zephyr 3B

Stability AI

S
3BChat / LLMQ4_K_M2.1GB

Rocket 3B

Pansophic

S
3BChat / LLMQ4_K_M2.1GB

StarCoder2 3B

BigCode

S
3BCodingQ4_K_M2.3GB

Stable Code 3B

Stability AI

S
3BCodingQ4_K_M2.1GB

PaliGemma 3B

Google

S
3BMultimodalQ4_K_M2.5GB

Gemma 2 2B

Google

S
2.6BChat / LLMQ4_K_M2.1GB

EXAONE 3.5 2.4B

LG AI

S
2.4BChat / LLMQ4_K_M2.0GB

Qwen2-VL 2B

Alibaba

S
2.2BMultimodalQ8_02.0GB

CodeGemma 2B

Google

S
2BCodingQ4_K_M2.0GB

MiniCPM-V 2.6

OpenBMB

S
2BMultimodalQ4_K_M2.1GB

Granite 3.3 2B

IBM

S
2BChat / LLMQ4_K_M1.9GB

Moondream 2

Moondream

S
1.8BMultimodalQ4_K_M1.5GB

SmolLM2 1.7B

HuggingFace

S
1.7BChat / LLMQ8_02.2GB

Qwen 2.5 1.5B

Alibaba

S
1.5BChat / LLMQ8_02.3GB

DeepSeek R1 Distill 1.5B

DeepSeek

S
1.5BChat / LLMQ8_02.3GB

Qwen 2.5 Coder 1.5B

Alibaba

S
1.5BCodingQ8_02.3GB

Yi Coder 1.5B

01.AI

S
1.5BCodingQ8_02.0GB

DeepSeek Coder 1.3B

DeepSeek

S
1.3BCodingQ8_01.8GB

Llama 3.2 1B Instruct

Meta

S
1.24BChat / LLMQ8_01.7GB

TinyLlama 1.1B

TinyLlama

S
1.1BChat / LLMQ8_01.6GB

Gemma 3 1B

Google

S
1BChat / LLMQ8_01.5GB

Falcon 3 1B

TII

S
1BChat / LLMQ8_02.2GB

Stable Diffusion 2.1 Base (CoreML)

Stability AI / Apple

S
0.86BImage GenerationCoreML-Palettized1.6GB

Stable Diffusion 1.5 (CoreML)

Runway

S
0.86BImage GenerationCoreML-Palettized2.5GB

Stable Diffusion 1.5 (GGUF)

Runway / GPUStack

S
0.86BImage GenerationQ8_02.3GB

Stable Diffusion 2.1 (GGUF)

Stability AI

S
0.86BImage GenerationQ8_02.7GB

Whisper Large v3 Turbo

OpenAI

S
0.81BSpeech RecognitionQ8_02.0GB

Whisper Medium

OpenAI

S
0.77BSpeech RecognitionQ8_01.9GB

Distil-Whisper Large v3

HuggingFace

S
0.76BSpeech RecognitionQ8_01.9GB

BGE Reranker v2 M3

BAAI

S
0.568BRerankerFP161.6GB

Qwen 2.5 0.5B

Alibaba

S
0.5BChat / LLMQ8_01.1GB

Qwen 2.5 Coder 0.5B

Alibaba

S
0.5BCodingQ8_01.1GB

Danube 3 500M

H2O.ai

S
0.5BChat / LLMQ8_01.0GB

SmolLM2 360M

HuggingFace

S
0.36BChat / LLMQ8_00.9GB

BGE Large EN v1.5

BAAI

S
0.335BEmbeddingFP161.1GB

MusicGen Small

Meta

S
0.3BAudio GenerationONNX-Q4F160.8GB

Whisper Small

OpenAI

S
0.24BSpeech RecognitionQ8_00.9GB

Nomic Embed Text v1.5

Nomic AI

S
0.137BEmbeddingFP160.8GB

SmolLM2 135M

HuggingFace

S
0.135BChat / LLMFP160.8GB

Kokoro 82M TTS

Kokoro

S
0.082BText to SpeechONNX-Q8F160.6GB

Whisper Base

OpenAI

S
0.074BSpeech RecognitionQ8_00.3GB

Whisper Base English

OpenAI

S
0.074BSpeech RecognitionQ8_00.3GB

Whisper Tiny English (Quantized)

OpenAI

S
0.039BSpeech RecognitionQ5_10.1GB

Whisper Tiny

OpenAI

S
0.039BSpeech RecognitionQ8_00.2GB

BGE Small EN v1.5

BAAI

S
0.033BEmbeddingQ8_00.1GB

Snowflake Arctic Embed S

Snowflake

S
0.033BEmbeddingQ8_00.1GB

Jina Reranker Tiny EN

Jina AI

S
0.033BRerankerFP160.1GB

all-MiniLM-L6-v2

Sentence Transformers

S
0.023BEmbeddingQ8_00.1GB

Piper TTS - Amy (English)

Rhasspy

S
0.02BText to SpeechONNX0.1GB

Piper TTS - Lessac (English)

Rhasspy

S
0.02BText to SpeechONNX0.1GB

Piper TTS - LibriTTS-R (English)

Rhasspy

S
0.02BText to SpeechONNX0.6GB

Piper TTS - Spanish (MLS)

Rhasspy

S
0.02BText to SpeechONNX0.1GB

Piper TTS - French (Siwis)

Rhasspy

S
0.02BText to SpeechONNX0.5GB

Piper TTS - German (Thorsten)

Rhasspy

S
0.02BText to SpeechONNX0.1GB

Piper TTS - Chinese (Huayan)

Rhasspy

S
0.02BText to SpeechONNX0.1GB

Piper TTS - Japanese (Kokoro)

Rhasspy

S
0.02BText to SpeechONNX0.1GB

Piper TTS - Korean

Rhasspy

S
0.02BText to SpeechONNX0.1GB

Piper TTS - Russian (Irina)

Rhasspy

S
0.02BText to SpeechONNX0.1GB

Piper TTS - Portuguese (Faber)

Rhasspy

S
0.02BText to SpeechONNX0.1GB

Piper TTS - Italian (Riccardo)

Rhasspy

S
0.02BText to SpeechONNX0.5GB

Piper TTS - Arabic (Kareem)

Rhasspy

S
0.02BText to SpeechONNX0.1GB
A

Great (6 models)

Runs well with good performance

Phi-3.5 Vision

Microsoft

A
4.2BMultimodalQ4_K_M3.2GB

Gemma 3 4B

Google

A
4BChat / LLMQ4_K_M2.8GB

Nemotron Mini 4B

NVIDIA

A
4BChat / LLMQ4_K_M3.0GB

Danube 3 4B

H2O.ai

A
4BChat / LLMQ4_K_M2.7GB

Phi-3.5 Mini 3.8B

Microsoft

A
3.8BChat / LLMQ5_K_M3.1GB

Phi-4 Mini 3.8B

Microsoft

A
3.8BChat / LLMQ4_K_M2.8GB
B

Good (3 models)

Runs but may be tight on memory

Yi 1.5 6B Chat

01.AI

B
6BChat / LLMQ4_K_M3.9GB

Stable Diffusion XL (CoreML)

Stability AI

B
3.5BImage GenerationCoreML3.3GB

Whisper Large v3

OpenAI

B
1.55BSpeech RecognitionQ8_03.4GB
C

Possible (12 models)

Needs partial CPU offloading, slower performance

EXAONE 3.5 7.8B

LG AI

C
7.8BChat / LLMQ4_K_M4.9GB

InternLM 2.5 7B

Shanghai AI Lab

C
7.7BChat / LLMQ4_K_M4.9GB

Qwen 2.5 Coder 7B

Alibaba

C
7.6BCodingQ4_K_M4.9GB

Mistral 7B Instruct v0.3

Mistral AI

C
7.3BChat / LLMQ4_K_M4.6GB

LLaVA 1.6 7B

LLaVA

C
7BMultimodalQ4_K_M5.0GB

Falcon 3 7B

TII

C
7BChat / LLMQ4_K_M5.0GB

OLMo 2 7B

Allen AI

C
7BChat / LLMQ4_K_M4.7GB

OpenChat 3.5 7B

OpenChat

C
7BChat / LLMQ4_K_M4.6GB

StarCoder2 7B

BigCode

C
7BCodingQ4_K_M4.7GB

Code Llama 7B

Meta

C
7BCodingQ4_K_M4.3GB

DeepSeek Coder 6.7B

DeepSeek

C
6.7BCodingQ4_K_M4.3GB

SDXL Turbo (GGUF)

Stability AI

C
3.5BImage GenerationQ5_05.0GB
D

Struggling (13 models)

Heavy offloading required, very slow

Code Llama 13B Instruct

Meta

D
13BCodingQ4_K_M7.8GB

Gemma 3 12B

Google

D
12BChat / LLMQ4_K_M7.3GB

Mistral Nemo 12B

Mistral AI

D
12BChat / LLMQ4_K_M7.5GB

Solar 10.7B

Upstage

D
10.7BChat / LLMQ4_K_M6.5GB

Falcon 3 10B

TII

D
10BChat / LLMQ4_K_M6.4GB

Gemma 2 9B Instruct

Google

D
9.2BChat / LLMQ5_K_M6.7GB

Yi 1.5 9B Chat

01.AI

D
9BChat / LLMQ4_K_M5.5GB

Yi Coder 9B

01.AI

D
9BCodingQ4_K_M5.5GB

CodeGemma 7B

Google

D
8.5BCodingQ4_K_M5.5GB

DeepSeek R1 Distill 8B

DeepSeek

D
8BChat / LLMQ5_K_M5.8GB

Llama 3.1 8B Instruct

Meta

D
8BChat / LLMQ5_K_M5.8GB

Granite 3.3 8B

IBM

D
8BChat / LLMQ4_K_M5.1GB

Qwen 2.5 7B Instruct

Alibaba

D
7.6BChat / LLMQ5_K_M6.2GB
F

Cannot Run (10 models)

Insufficient hardware to run this model

Llama 3.1 70B Instruct

Meta

F
70BChat / LLMFP16142.0GB

Qwen 2.5 32B

Alibaba

F
32BChat / LLMQ4_K_M19.0GB

Gemma 3 27B

Google

F
27BChat / LLMQ4_K_M15.9GB

Mistral Small 22B

Mistral AI

F
22BChat / LLMQ4_K_M12.9GB

Phi-4

Microsoft

F
14BChat / LLMQ8_015.0GB

Qwen 2.5 14B

Alibaba

F
14BChat / LLMQ8_015.1GB

Qwen 2.5 Coder 14B

Alibaba

F
14BCodingQ8_015.1GB

FLUX.1 Schnell (GGUF)

Black Forest Labs

F
12BImage GenerationQ5_014.0GB

FLUX.1 Dev (GGUF)

Black Forest Labs

F
12BImage GenerationQ5_014.0GB

Stable Diffusion 3 Medium (GGUF)

Stability AI

F
2.5BImage GenerationQ8_09.2GB

Similar GPUs

NVIDIA GeForce GTX 1050 Ti

4GB$139nvidia

NVIDIA GeForce GTX 1650 SUPER

4GB$159nvidia

NVIDIA GeForce GTX 1060 6GB

6GB$249nvidia

NVIDIA GeForce GTX 1660

6GB$219nvidia

NVIDIA GeForce GTX 1660 SUPER

6GB$229nvidia

NVIDIA GeForce GTX 1660 Ti

6GB$279nvidia

NVIDIA GeForce RTX 2060

6GB$349nvidia

NVIDIA RTX 3060 Laptop

6GBnvidia

AMD Radeon RX 6600

8GB$329amd

AMD Radeon RX 6600 XT

8GB$379amd

AMD Radeon RX 7600

8GB$269amd

Apple Silicon (Unknown)

8GBapple

Frequently Asked Questions

What AI models can I run on the NVIDIA GeForce GTX 1650?
The NVIDIA GeForce GTX 1650 with 4GB VRAM can run 74 AI models smoothly (grade B or better), including Llama 3.2 3B Instruct, Qwen 2.5 3B, Qwen 2.5 Coder 3B and 71 more.
How much VRAM does the NVIDIA GeForce GTX 1650 have?
The NVIDIA GeForce GTX 1650 has 4GB of VRAM, which is sufficient for running many local AI models.
Is the NVIDIA GeForce GTX 1650 good for AI?
With 4GB VRAM, the NVIDIA GeForce GTX 1650 can run 74 out of 109 models in our database at grade B or better. You may need smaller quantizations for larger models.