Llemma 7B — Hardware Requirements & GPU Compatibility
ChatMathReasoning
Llemma 7B is a 7B-parameter open language model from EleutherAI. It supports a context window of up to 4,096 tokens. At Q4_K_M it needs about 5.57 GB of VRAM — see which GPUs and Macs can run it below.
Runs great on a 24 GB GPU
Runs from 8 GB (Q4_K_M · 4K)Best on 24 GB (Q4_K_M · ~118 tok/s · 4K)Macs from 16 GB (Q4_K_M · ~25 tok/s)
Llama 2 Community
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Llemma 7B
One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.
Recommended
vllm serve EleutherAI/llemma_7b --max-model-len 4096vLLM repo with its own quantisation
Download: Weights on Hugging Face
Ways to runOther ways
each opens step-by-stepRun in cloud
Runs on RTX 3080 Ti (12 GB) (5 GB headroom) · Q4_K_M
- Generation speed
- ~107 tok/s
- generation speed
- Cost per 1M output tokens
- $0.32
- per 1M output tokens
How Much VRAM Does Llemma 7B Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 4.3 GB | — | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 4K context · ~151 tok/s | |
| 4.8 GB | — | 3-bit medium — 3-bit medium quantization | Fits: 4K context · ~137 tok/s | |
| 5.6 GB | — | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 4K context · ~118 tok/s | |
| 6.4 GB | — | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Fits: 4K context · ~103 tok/s | |
| 7.2 GB | — | 6-bit, very good — 6-bit quantization, very good quality | Fits: 4K context · ~92 tok/s | |
| 8.4 GB | — | 8-bit, near-lossless — 8-bit quantization, near-lossless | Fits: 4K context · ~78 tok/s | |
| 15.4 GB | — | 16-bit · bfloat — Brain floating point 16 — preferred for training | Fits: 4K context · ~43 tok/s |
Selected Q4_K_M
4-bit medium quantization6.6GBat 4K context
- Weights
- 4.2 GB
- KV cache
- 2.1 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · 17.4 GB free · ~118 tok/s
≥50 tok/s needs ≥428 GB/s · RTX 5090 ~209, RTX 3090 Ti ~118
Which Hardware Can Run Llemma 7B?
Q4_K_M · 5.6 GBLlemma 7B (Q4_K_M) requires 5.6 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 1.1 GB, bringing total usage to 6.7 GB. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti, NVIDIA GeForce RTX 3070 Ti. 151 devices with unified memory can run Llemma 7B, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), NVIDIA Jetson Orin Nano 8GB (Super).
≤ 11 GB
≤ 11 GB tier →fits · 1.4 GB free · up to 4K ctx
16 GPUs · 1 device family
RTX 3080~89Simulate output on RTX 3080RTX 2080 Ti~72Simulate output on RTX 2080 TiRTX 3070 Ti~71Simulate output on RTX 3070 TiOrin Nano~12Simulate output on NVIDIA Jetson Orin Nano 8GB (Super)20 GB
fits · 13.4 GB free · up to 4K ctx
1 GPU
RX 7900 XT~86Simulate output on RX 7900 XT32–40 GB
fits · 25.4 GB free · up to 4K ctx
1 GPU · 3 Mac classes · 4 device families
RTX 5090~209Simulate output on RTX 5090M5 Max36 GB~58Simulate output on Mac Studio M5 Max (36 GB)M532 GB~19Simulate output on MacBook Air 13" M5 (32 GB)Orin~24Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~16Simulate output on Snapdragon X Elite Copilot+ PC
By VRAM tier · typical GPUs and Macs at Q4_K_M
How good is Llemma 7B?
llmrun Score blends public benchmark boards into one number; 100 = the average open model.
Not enough data yet for an llmrun Score.
A Score needs results on enough independent boards. How the Composite Score is built
Benchmarks
This is the single board showing results for this model; badges show who ran the test.
- MMLU-Pro
TIGER-LabIndependent
Independent23.4%
#128/146 open
Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Olmo 3 7B InstructAllen AI | — | 8.4 GB | ~114 | same size |
| Mistral 7B Instruct v0.3Mistral AI | — | 6.8 GB | ~133 | same size |
| Phi 3 Small 8k InstructMicrosoft | — | 16.2 GB | ~43 | same size |
| CodeQwen1.5 7BAlibaba | — | 5.7 GB | ~137 | same size |
| Llava 1.5 7B HFllava-hf | — | 5 GB | ~141 | same size |
| Baichuan2 7B Basebaichuan-inc | — | 5 GB | ~142 | same size |
Specifications
- Publisher
- EleutherAI
- Parameters
- 7B
- Architecture
- LlamaForCausalLM
- Context Length
- 4,096 tokens
- Vocabulary Size
- 32,016
- Release Date
- 2023-09-12
- License
- Llama 2 Community
Frequently Asked Questions
How much VRAM does Llemma 7B need?
Llemma 7B requires 5.6 GB of VRAM at Q4_K_M, or 15.4 GB at BF16. Full 4K context adds up to 1.1 GB (6.7 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 7B × 4.8 bits ÷ 8 = 4.2 GB
KV Cache + Overhead ≈ 1.4 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 ≈ 2.5 GB (at full 4K context)
VRAM usage by quantization
What's the best quantization for Llemma 7B?
For Llemma 7B, Q4_K_M (5.6 GB) offers the best balance of quality and VRAM usage. Q5_K_M (6.4 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 4.3 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run Llemma 7B on a Mac?
Yes — Mac Mini M2 Pro (16 GB) and 133 other Macs can run Llemma 7B. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 5.6 GB of usable unified memory (RAM minus macOS overhead).
Can I run Llemma 7B locally?
Yes — Llemma 7B can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Llemma 7B?
At Q4_K_M, Llemma 7B can reach ~862 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~118 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 ÷ 5.57 × 0.65 = ~934 tok/s
Estimated speed at Q4_K_M (5.6 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Llemma 7B?
At Q4_K_M, the download is about 4.20 GB. The full-precision BF16 version is 14.00 GB. The smallest option (Q2_K) is 2.98 GB.
Which GPUs can run Llemma 7B?
52 consumer GPUs can run Llemma 7B 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. 40 GPUs have plenty of headroom for comfortable inference.
Which devices can run Llemma 7B?
151 devices with unified memory can run Llemma 7B at Q4_K_M (5.6 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (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.