FITMYLLM · JULY 28, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
46.7B
Parameters (13B active)
Quantization Options Select your GPU for speed estimates Context length: 4K 8K 16K 32K
Quant Bits VRAM @ 16K Quality IQ3_XXS 3.25 20.0 GB
19.5 + 0.5 KV
low IQ3_XS 3.5 21.4 GB
20.9 + 0.5 KV
low Q3_K_S 3.64 22.2 GB
21.7 + 0.5 KV
low IQ3_M 3.76 22.9 GB
22.4 + 0.5 KV
low Q3_K_M 4 24.3 GB
23.8 + 0.5 KV
low Q3_K_L 4.3 26.1 GB
25.6 + 0.5 KV
moderate IQ4_XS 4.46 27.0 GB
26.5 + 0.5 KV
moderate Q4_K_S 4.67 28.2 GB
27.7 + 0.5 KV
moderate Q4_K_M 4.89 29.5 GB
29.0 + 0.5 KV
good Q5_K_S 5.57 33.5 GB
33.0 + 0.5 KV
good Q5_K_M 5.7 34.3 GB
33.8 + 0.5 KV
good Q6_K 6.56 39.3 GB
38.8 + 0.5 KV
excellent Q8_0 8.5 50.6 GB
50.1 + 0.5 KV
lossless FP16 16 94.4 GB
93.9 + 0.5 KV
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 NOUS-HERMES-2-MIXTRAL-8X7B-DPO 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 IQ3_XXS — 19.5 GB VRAM IQ3_XS — 20.9 GB VRAM Q3_K_S — 21.7 GB VRAM IQ3_M — 22.4 GB VRAM Q3_K_M — 23.8 GB VRAM Q3_K_L — 25.6 GB VRAM IQ4_XS — 26.5 GB VRAM Q4_K_S — 27.7 GB VRAM Q4_K_M — 29.0 GB VRAM Q5_K_S — 33.0 GB VRAM Q5_K_M — 33.8 GB VRAM Q6_K — 38.8 GB VRAM Q8_0 — 50.1 GB VRAM FP16 — 93.9 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 nous-hermes2-mixtral:8x7b-dpo-q4_K_MCOPY
Downloads and runs automatically. Add --verbose for speed stats.
▸ 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 Nous-Hermes-2-Mixtral-8x7B-DPO
Build Hardware for Nous-Hermes-2-Mixtral-8x7B-DPO Nous Hermes 2 Mixtral — community fine-tune of Mixtral with improved helpfulness.
Read full model card ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
Nous-Hermes-2-Mixtral-8x7B-DPO — 46.7B MoE. ▸ SPECIFICATIONS
PARAMETERS 46.7B (13B active)
ARCHITECTURE Mixture of Experts
CONTEXT LENGTH 32K tokens
CAPABILITIES chat
RELEASE DATE 2024-01-11
PROVIDER NousResearch
FAMILY mistral ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ3_XXS 3.25 19.5 GB 82% IQ3_XS 3.5 20.9 GB 84% Q3_K_S 3.64 21.7 GB 85% IQ3_M 3.76 22.4 GB 86% Q3_K_M 4 23.8 GB 88% Q3_K_L 4.3 25.6 GB 90% IQ4_XS 4.46 26.5 GB 92% Q4_K_S 4.67 27.7 GB 93% Q4_K_M 4.89 29.0 GB 94% Q5_K_S 5.57 33.0 GB 96% Q5_K_M 5.7 33.8 GB 96% Q6_K 6.56 38.8 GB 97% Q8_0 8.5 50.1 GB 100% FP16 16 93.9 GB 100%
§ 01 BENCHMARK SCORES
MMLU-PRO 29.6
MATH 12.2
IFEval 59.0
BBH 37.1
GPQA 9.5
MUSR 16.7
Arena Elo 1099.0
§ 02 RUN COMMAND
Run Nous-Hermes-2-Mixtral-8x7B-DPO locally with Ollama — needs 29.0 GB VRAM at Q4_K_M:
$ ollama run nous-hermes2-mixtral:8x7b
§ 03 COMPATIBLE GPUs
30 @ Q4_K_M Feedback 