Can I run Llama 3.1 70B (lorablated) on my device?

Llama 3.1 70B (lorablated) requires a minimum of 42.5GB VRAM. Use RunThisModel to check your specific hardware compatibility and find the best quantization for your device.

How much VRAM does Llama 3.1 70B (lorablated) need?

Llama 3.1 70B (lorablated) needs 42.5GB VRAM at minimum (BF16 quantization). Higher quality quantizations need more: BF16: 140.5GB, Q4_K_M: 42.5GB.

How do I download Llama 3.1 70B (lorablated)?

You can download Llama 3.1 70B (lorablated) in GGUF format from HuggingFace (42GB minimum). Use the RunThisModel iOS app to download and run it directly on your device, or download manually from HuggingFace.

Can Llama 3.1 70B (lorablated) run on iPhone?

Llama 3.1 70B (lorablated) at 70B parameters is too large for most iPhones. Consider using an iPad with M-series chip or Mac with Apple Silicon.

mlabonne

Llama 3.1 70B (lorablated)

Name: Llama 3.1 70B (lorablated)
Author: mlabonne

Llama-3.1-70B-Instruct with abliteration applied via LoRA merge. Cleanest 70B refusal-removed pick — keeps the official Instruct quality.

70B parametersllamallama3.1128K context42.5GB - 140.5GB VRAM

Check Your Hardware

See which quantizations of Llama 3.1 70B (lorablated) your hardware can run.

Quantization Options

Quantization	Bits	File Size	VRAM Needed	RAM Needed	Quality
BF16	16	140 GB	140.5 GB	141 GB	100%
Q4_K_M	4.5	42 GB	42.5 GB	43 GB	85%

Context window & KV cache

Adds 2.50 GB to VRAM

Long chats and RAG inputs cost real memory. Drag to see how 32K vs 128K context shifts your grade.

Model native max: 128K tokens. KV-cache estimate is approximate (±30 %); real usage depends on attention layout.

How to run Llama 3.1 70B (lorablated)

Pick a runtime — copy & paste. Commands are pre-filled with this model’s repo.

GUI. Browse → download → chat. MLX on Apple Silicon.

LM Studio home →

1
Open LM Studio
Go to the 🔍 Search tab.

Search for

bartowski/Llama-3.1-70B-Instruct-lorablated-GGUF

3
Download
Pick the Q4_K_M quant — best balance of size vs. quality.
4
Chat
Hit ▶ Load Model and start chatting. Toggle 'Local Server' to expose an OpenAI-compatible API on :1234.

Community benchmarks

Real tokens/sec reports from people running Llama 3.1 70B (lorablated) on actual hardware.

No community runs yet for this model. Be the first to submit your numbers.

Self-host serving plan

Want to host Llama 3.1 70B (lorablated)for many users? Or run it on a card that’s technically too small? Slide the knobs.

Concurrent users: 1VRAM budget: 24 GBSystem RAM: 64 GBAllow NVMe spilloverPermit weights to spill to NVMe SSD (last resort, ~50× slower)

VRAM needed

45.1 GB

42.5 GB weights + 2.1 GB KV

Aggregate tok/s

across 1 user

Per-user tok/s

70 B dense

⚠ Will spill 21.1 GB of weights to system RAM (~5× slower per offloaded layer). Use llama.cpp’s --cpu-offload-gb or vLLM’s --swap-space.

Throughput is a sub-linear estimate: doubling users adds ~70 % of single-user TPS until ~8, then plateaus on memory bandwidth. MoE models scale concurrency much better because each user activates a different subset of experts.

Download & Run

HuggingFace

View model & download weights

Ollama

One-command install & run

See It In Action

Real model outputs generated via RunThisModel.com — watch responses stream in real time.

Llama 3.3 70B responding...

Outputs generated by real AI models via RunThisModel.com. Generation speed shown is from cloud inference. Local speeds vary by hardware — check your device.

Frequently Asked Questions

How much VRAM do I need to run Llama 3.1 70B (lorablated)?

Llama 3.1 70B (lorablated) requires 42.5GB VRAM minimum with BF16 quantization. For full precision, you need 140.5GB VRAM.

What is the best quantization for Llama 3.1 70B (lorablated)?

Q4_K_M offers the best balance of quality and VRAM usage. Q8_0 is near-lossless if you have enough VRAM.