FITMYLLM · AUGUST 17, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
119B
Parameters (6.5B active)
Quantization Options Select your GPU for speed estimates Context length: 4K 8K 16K 32K 64K 128K 250K
Quant Bits VRAM @ 16K Quality IQ2_XXS 2.38 36.2 GB
35.9 + 0.3 KV
low IQ2_M 2.93 44.3 GB
44.1 + 0.3 KV
low Q2_K 3.16 47.8 GB
47.5 + 0.3 KV
low IQ3_XXS 3.25 49.1 GB
48.8 + 0.3 KV
low IQ3_XS 3.5 52.8 GB
52.6 + 0.3 KV
low Q3_K_S 3.64 54.9 GB
54.6 + 0.3 KV
low IQ3_M 3.76 56.7 GB
56.4 + 0.3 KV
low Q3_K_M 4 60.3 GB
60.0 + 0.3 KV
low Q3_K_L 4.3 64.7 GB
64.5 + 0.3 KV
moderate IQ4_XS 4.46 67.1 GB
66.8 + 0.3 KV
moderate Q4_K_S 4.67 70.2 GB
70.0 + 0.3 KV
moderate Q4_K_M 4.89 73.5 GB
73.2 + 0.3 KV
good Q5_K_S 5.57 83.6 GB
83.3 + 0.3 KV
good Q5_K_M 5.7 85.5 GB
85.3 + 0.3 KV
good Q6_K 6.56 98.3 GB
98.1 + 0.3 KV
excellent Q8_0 8.5 127.2 GB
126.9 + 0.3 KV
lossless FP16 16 238.8 GB
238.5 + 0.3 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 MISTRAL SMALL 4 119B 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_XXS — 35.9 GB VRAM IQ2_M — 44.1 GB VRAM Q2_K — 47.5 GB VRAM IQ3_XXS — 48.8 GB VRAM IQ3_XS — 52.6 GB VRAM Q3_K_S — 54.6 GB VRAM IQ3_M — 56.4 GB VRAM Q3_K_M — 60.0 GB VRAM Q3_K_L — 64.5 GB VRAM IQ4_XS — 66.8 GB VRAM Q4_K_S — 70.0 GB VRAM Q4_K_M — 73.2 GB VRAM Q5_K_S — 83.3 GB VRAM Q5_K_M — 85.3 GB VRAM Q6_K — 98.1 GB VRAM Q8_0 — 126.9 GB VRAM FP16 — 238.5 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 mistral:119b-q4_K_MCOPY
Tag may need adjustment — check ollama.com/library/mistral 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 Mistral Small 4 119B
Build Hardware for Mistral Small 4 119B Mistral Small 4 — 119B MoE with vision, coding, math. Very efficient.
Read full model card ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
Mistral Small 4 119B — 119B MoE. ▸ SPECIFICATIONS
PARAMETERS 119B (6.5B active)
ARCHITECTURE Mixture of Experts
CONTEXT LENGTH 250K tokens
CAPABILITIES chat, coding, reasoning, math, vision, tool_use
RELEASE DATE 2026-03-16
PROVIDER Mistral AI
FAMILY mistral ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_XXS 2.38 35.9 GB 65% IQ2_M 2.93 44.1 GB 75% Q2_K 3.16 47.5 GB 78% IQ3_XXS 3.25 48.8 GB 82% IQ3_XS 3.5 52.6 GB 84% Q3_K_S 3.64 54.6 GB 85% IQ3_M 3.76 56.4 GB 86% Q3_K_M 4 60.0 GB 88% Q3_K_L 4.3 64.5 GB 90% IQ4_XS 4.46 66.8 GB 92% Q4_K_S 4.67 70.0 GB 93% Q4_K_M 4.89 73.2 GB 94% Q5_K_S 5.57 83.3 GB 96% Q5_K_M 5.7 85.3 GB 96% Q6_K 6.56 98.1 GB 97% Q8_0 8.5 126.9 GB 100% FP16 16 238.5 GB 100%
§ 01 BENCHMARK SCORES
MMLU-PRO 50.7
MATH 49.5
IFEval 84.0
BBH 52.7
GPQA 24.9
MUSR 17.2
Arena Elo 1480.0
GPQA Diamond 71.2
AIME 83.8
LiveCodeBench 63.6
HLE 4.3
§ 03 COMPATIBLE GPUs
30 @ Q4_K_M Feedback 