CodeLlama 34B Instruct GGUF — Hardware Requirements & GPU Compatibility
ChatCode
2.2K downloads
108 likesSpecifications
- Publisher
- TheBloke
- Family
- Llama 3
- Parameters
- 34B
- Release Date
- 2023-09-27
- License
- Llama 2 Community
Get Started
HuggingFace
How Much VRAM Does CodeLlama 34B Instruct GGUF Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| Q2_K | 3.40 | 15.9 GB | — | 14.45 GB | 2-bit quantization with K-quant improvements |
| Q3_K_S | 3.50 | 16.4 GB | — | 14.88 GB | 3-bit small quantization |
| Q3_K_M | 3.90 | 18.2 GB | — | 16.57 GB | 3-bit medium quantization |
| Q4_0 | 4.00 | 18.7 GB | — | 17.00 GB | 4-bit legacy quantization |
| Q4_K_M | 4.80 | 22.4 GB | — | 20.40 GB | 4-bit medium quantization — most popular sweet spot |
| Q5_K_M | 5.70 | 26.6 GB | — | 24.23 GB | 5-bit medium quantization — good quality/size tradeoff |
| Q6_K | 6.60 | 30.9 GB | — | 28.05 GB | 6-bit quantization, very good quality |
| Q8_0 | 8.00 | 37.4 GB | — | 34.00 GB | 8-bit quantization, near-lossless |
Which GPUs Can Run CodeLlama 34B Instruct GGUF?
Q4_K_M · 22.4 GBCodeLlama 34B Instruct GGUF (Q4_K_M) requires 22.4 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 30+ GB is recommended. 5 GPUs can run it, including NVIDIA GeForce RTX 5090.
All compatible consumer-level GPUs are running near their VRAM limit. You may also want to consider professional GPUs (e.g., NVIDIA A100, H100) which offer significantly more VRAM. For more headroom and better throughput, consider a multi-GPU configuration with tensor parallelism (supported by tools like vLLM, llama.cpp, or text-generation-inference).
Which Devices Can Run CodeLlama 34B Instruct GGUF?
Q4_K_M · 22.4 GB21 devices with unified memory can run CodeLlama 34B Instruct GGUF, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, Mac Mini M4 Pro (24 GB).
Runs great
— Plenty of headroomDecent
— Enough memory, may be tightRelated Models
Frequently Asked Questions
- How much VRAM does CodeLlama 34B Instruct GGUF need?
CodeLlama 34B Instruct GGUF requires 22.4 GB of VRAM at Q4_K_M, or 37.4 GB at Q8_0.
VRAM = Weights + KV Cache + Overhead
Weights = 34B × 4.8 bits ÷ 8 = 20.4 GB
KV Cache + Overhead ≈ 2 GB (at 2K context + ~0.3 GB framework)
VRAM usage by quantization
Q4_K_M22.4 GB- Can NVIDIA GeForce RTX 4090 run CodeLlama 34B Instruct GGUF?
Yes, at Q5_0 (23.4 GB) or lower. Higher quantizations like Q5_K_S (25.7 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.
- What's the best quantization for CodeLlama 34B Instruct GGUF?
For CodeLlama 34B Instruct GGUF, Q4_K_M (22.4 GB) offers the best balance of quality and VRAM usage. Q5_0 (23.4 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 15.9 GB.
VRAM requirement by quantization
Q2_K15.9 GB~75%Q4_018.7 GB~85%Q4_K_M ★22.4 GB~89%Q5_023.4 GB~90%Q5_K_S25.7 GB~92%Q8_037.4 GB~99%★ Recommended — best balance of quality and VRAM usage.
- Can I run CodeLlama 34B Instruct GGUF on a Mac?
CodeLlama 34B Instruct GGUF requires at least 15.9 GB at Q2_K, 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 CodeLlama 34B Instruct GGUF locally?
Yes — CodeLlama 34B Instruct GGUF can run locally on consumer hardware. At Q4_K_M quantization it needs 22.4 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is CodeLlama 34B Instruct GGUF?
At Q4_K_M, CodeLlama 34B Instruct GGUF can reach ~130 tok/s on AMD Instinct MI300X. On NVIDIA GeForce RTX 4090: ~29 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: AMD Instinct MI300X → 5300 ÷ 22.4 × 0.55 = ~130 tok/s
Estimated speed at Q4_K_M (22.4 GB)
AMD Instinct MI300X~130 tok/sNVIDIA GeForce RTX 4090~29 tok/sNVIDIA H100 SXM~97 tok/sAMD Instinct MI250X~80 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of CodeLlama 34B Instruct GGUF?
At Q4_K_M, the download is about 20.40 GB. The full-precision Q8_0 version is 34.00 GB. The smallest option (Q2_K) is 14.45 GB.
- Which GPUs can run CodeLlama 34B Instruct GGUF?
5 consumer GPUs can run CodeLlama 34B Instruct GGUF at Q4_K_M (22.4 GB). Top options include AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090.
- Which devices can run CodeLlama 34B Instruct GGUF?
21 devices with unified memory can run CodeLlama 34B Instruct GGUF at Q4_K_M (22.4 GB), including Mac Mini M4 (32 GB), Mac Mini M4 Pro (24 GB), Mac Mini M4 Pro (48 GB), Mac Pro M2 Ultra (192 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.