utQwen · Alibaba · Downloadable
Qwen3-30B-A3B-MLX
Generates text for conversations and writing.
- Source listed
- Jun 11, 2025
- Parameters
- 30B
- 4-bit weights only
- ≈ 15 GB
- License
- apache-2.0
YOUR MEMORY BUDGET
Local AI models for 24GB VRAM.
Explore models by theoretical weight storage. This shortlist covers available downloadable language models with a known or name-derived parameter count.
22 GBleft for weights at 4-bit
24 GB budget − 2 GB reserve
24 GB budget − 2 GB reserve
The reserve is your assumption, not a measured overhead allowance. The default 2 GB may be too little for your runtime or context length. Results are not confirmed fits and have no verified speed rating. Weight estimates use decimal GB and do not include quantization overhead.
405 models within the weight budget
Largest parameter count first · not a quality rankingQwen · Alibaba · Downloadable
Qwen3-30B-A3B-Thinking-2507
Generates text for conversations and writing. Supports a reasoning mode.
- Source listed
- Jul 29, 2025
- Parameters
- 30B
- 4-bit weights only
- ≈ 15 GB
- License
- apache-2.0
Qwen · Alibaba · Downloadable
Qwen3-Coder-30B-A3B-Instruct
Designed for coding and software tasks. Generates text for conversations and writing.
- Source listed
- Jul 31, 2025
- Parameters
- 30B
- 4-bit weights only
- ≈ 15 GB
- License
- apache-2.0
Qwen · Alibaba · Downloadable
Qwen3-VL-30B-A3B-Instruct
Understands text and images and responds in text.
- Source listed
- Sep 30, 2025
- Parameters
- 30B
- 4-bit weights only
- ≈ 15 GB
- License
- apache-2.0
Qwen · Alibaba · Downloadable
Qwen3-VL-30B-A3B-Thinking
Understands text and images and responds in text. Supports a reasoning mode.
- Source listed
- Sep 30, 2025
- Parameters
- 30B
- 4-bit weights only
- ≈ 15 GB
- License
- apache-2.0
Google · Downloadable
gemma-3-27b-it
Understands text and images and responds in text.
- Source listed
- Mar 1, 2025
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- gemma
Google · Downloadable
gemma-3-27b-it-qat-q4_0
Understands text and images and responds in text.
- Source listed
- Mar 20, 2025
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- gemma
Google · Downloadable
gemma-3-27b-it-qat-q4_0-unquantized
Understands text and images and responds in text.
- Source listed
- Apr 15, 2025
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- gemma
Google · Downloadable
gemma-3-27b-pt
Understands text and images and responds in text.
- Source listed
- Mar 1, 2025
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- gemma
Google · Downloadable
gemma-3-27b-pt-qat-q4_0
Understands text and images and responds in text.
- Source listed
- Mar 20, 2025
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- gemma
Google · Downloadable
medgemma-27b-it
Understands text and images and responds in text.
- Source listed
- Jul 9, 2025
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- other
Google · Downloadable
medgemma-27b-text-it
Generates text for conversations and writing.
- Source listed
- May 19, 2025
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- other
Google · Downloadable
translategemma-27b-it
Understands text and images and responds in text.
- Source listed
- Jan 12, 2026
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- gemma
Google · Downloadable
txgemma-27b-chat
Generates text for conversations and writing.
- Source listed
- Mar 21, 2025
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- other
Google · Downloadable
txgemma-27b-predict
Generates text for conversations and writing.
- Source listed
- Mar 21, 2025
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- other
Microsoft · Downloadable
Fara1.5-27B
Understands text and images and responds in text.
- Source listed
- Jul 17, 2026
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- mit
NVIDIA · Downloadable
Qwen3.6-27B
Generates text for conversations and writing.
- Source listed
- Jun 22, 2026
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- apache-2.0
NVIDIA · Downloadable
Qwen3.8-27B
Generates text for conversations and writing.
- Source listed
- Sep 4, 2026
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- apache-2.0
Qwen · Alibaba · Downloadable
Qwen3.5-27B
Understands text and images and responds in text.
- Source listed
- Feb 24, 2026
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- apache-2.0
Qwen · Alibaba · Downloadable
Qwen3.6-27B
Understands text and images and responds in text.
- Source listed
- Apr 21, 2026
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- apache-2.0
Qwen · Alibaba · Downloadable
Qwen3.8-27B
Understands text and images and responds in text.
- Source listed
- Aug 5, 2026
- Parameters
- 27B
- 4-bit weights only
- ≈ 13.5 GB
- License
- apache-2.0
Google · Downloadable
diffusiongemma-26B-A4B-it
Understands text and images and responds in text.
- Source listed
- Jun 9, 2026
- Parameters
- 26B
- 4-bit weights only
- ≈ 13 GB
- License
- apache-2.0
Google · Downloadable
gemma-4-26B-A4B
Understands text and images and responds in text.
- Source listed
- Mar 12, 2026
- Parameters
- 26B
- 4-bit weights only
- ≈ 13 GB
- License
- apache-2.0
Google · Downloadable
gemma-4-26B-A4B-it
Understands text and images and responds in text.
- Source listed
- Mar 11, 2026
- Parameters
- 26B
- 4-bit weights only
- ≈ 13 GB
- License
- apache-2.0
What changes with a 24GB budget?
With 2 GB set aside, 22 GB remains for weights. At 4-bit that allows up to 44 billion parameters in the theoretical calculation. An 8-billion-parameter model alone needs approximately 4 GB at this precision, before its conversation cache and runtime.
Why might a listed model still fail to load?
Its actual weight format may need more space, the runtime may not support its architecture, or the conversation cache may exceed your reserve. Vision models can also need space for image processing. Check the original model card and measure your exact setup before relying on it.
Read the memory guide and technical sources · Compare two models