FITMYLLM · AUGUST 17, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
Quantization Options Select your GPU for speed estimates Quant Bits VRAM @ 2K Quality IQ2_M 2.93 24.4 GB
low Q2_K 3.16 26.2 GB
low IQ3_XXS 3.25 27.0 GB
low IQ3_XS 3.5 29.0 GB
low Q3_K_S 3.64 30.2 GB
low IQ3_M 3.76 31.1 GB
low Q3_K_M 4 33.1 GB
low Q3_K_L 4.3 35.5 GB
moderate IQ4_XS 4.46 36.8 GB
moderate Q4_K_S 4.67 38.5 GB
moderate Q4_K_M 4.89 40.3 GB
good Q5_K_S 5.57 45.9 GB
good Q5_K_M 5.7 46.9 GB
good Q6_K 6.56 54.0 GB
excellent Q8_0 8.5 69.8 GB
lossless FP16 16 130.9 GB
lossless
Select your GPU above to see speed estimates and compatibility for each quantization.
Deploying for a team or in production? Size GPUs, cost & scaling in Enterprise → ▸ READY TO RUN THIS? RENT BY THE HOUR
RENT A GPU AND RUN LLAMA 1 65B NOW
Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.
Community Ratings Chat Coding Reasoning Creative Vision Roleplay Agentic
Loading ratings...
Run this model IQ2_M — 24.4 GB VRAM Q2_K — 26.2 GB VRAM IQ3_XXS — 27.0 GB VRAM IQ3_XS — 29.0 GB VRAM Q3_K_S — 30.2 GB VRAM IQ3_M — 31.1 GB VRAM Q3_K_M — 33.1 GB VRAM Q3_K_L — 35.5 GB VRAM IQ4_XS — 36.8 GB VRAM Q4_K_S — 38.5 GB VRAM Q4_K_M — 40.3 GB VRAM Q5_K_S — 45.9 GB VRAM Q5_K_M — 46.9 GB VRAM Q6_K — 54.0 GB VRAM Q8_0 — 69.8 GB VRAM FP16 — 130.9 GB VRAM
Ollama llama.cpp vLLM LM Studio KoboldCpp Jan Docker
▸ Easiest way to get started · Beginners
DOCS ↗ curl -fsSL https://ollama.com/install.sh | shCOPY
$ ollama run llama:65b-q4_K_MCOPY
Tag may need adjustment — check ollama.com/library/llama for available tags.
▸ SETUP GUIDE >_
Auto-setup with fitmyllm CLI Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.
Auto-detect GPU Live tok/s in chat Speed benchmarks 9 inference engines
GPUs that can run this model At Q4_K_M quantization. Sorted by minimum VRAM.
Find the best GPU for LLaMA 1 65B
Build Hardware for LLaMA 1 65B LLaMA 1 65B — largest original LLaMA. Impressive for its time.
Read full model card ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
LLaMA 1 65B — 65.2B Dense. ▸ SPECIFICATIONS
PARAMETERS 65.2B
ARCHITECTURE Dense Transformer
CONTEXT LENGTH 2K tokens
CAPABILITIES chat
RELEASE DATE 2023-02-24
PROVIDER Meta
FAMILY llama ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_M 2.93 24.4 GB 75% Q2_K 3.16 26.2 GB 78% IQ3_XXS 3.25 27.0 GB 82% IQ3_XS 3.5 29.0 GB 84% Q3_K_S 3.64 30.2 GB 85% IQ3_M 3.76 31.1 GB 86% Q3_K_M 4 33.1 GB 88% Q3_K_L 4.3 35.5 GB 90% IQ4_XS 4.46 36.8 GB 92% Q4_K_S 4.67 38.5 GB 93% Q4_K_M 4.89 40.3 GB 94% Q5_K_S 5.57 45.9 GB 96% Q5_K_M 5.7 46.9 GB 96% Q6_K 6.56 54.0 GB 97% Q8_0 8.5 69.8 GB 100% FP16 16 130.9 GB 100%
§ 01 BENCHMARK SCORES
HumanEval 23.7
MMLU-PRO 48.1
MATH 44.1
IFEval 81.2
BBH 54.1
GPQA 24.6
MUSR 22.3
§ 03 COMPATIBLE GPUs
30 @ Q4_K_M Feedback 