FITMYLLM · AUGUST 17, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
Quantization Options Select your GPU for speed estimates Context length: 4K 8K 16K 32K 64K 128K
Quant Bits VRAM @ 16K Quality IQ2_XXS 2.38 41.5 GB
37.4 + 4.1 KV
low IQ2_M 2.93 50.0 GB
45.9 + 4.1 KV
low Q2_K 3.16 53.6 GB
49.5 + 4.1 KV
low IQ3_XXS 3.25 55.0 GB
50.9 + 4.1 KV
low IQ3_XS 3.5 58.9 GB
54.7 + 4.1 KV
low Q3_K_S 3.64 61.0 GB
56.9 + 4.1 KV
low IQ3_M 3.76 62.9 GB
58.8 + 4.1 KV
low Q3_K_M 4 66.6 GB
62.5 + 4.1 KV
low Q3_K_L 4.3 71.3 GB
67.1 + 4.1 KV
moderate IQ4_XS 4.46 73.7 GB
69.6 + 4.1 KV
moderate Q4_K_S 4.67 77.0 GB
72.9 + 4.1 KV
moderate Q4_K_M 4.89 80.4 GB
76.3 + 4.1 KV
good Q5_K_S 5.57 90.9 GB
86.8 + 4.1 KV
good Q5_K_M 5.7 93.0 GB
88.8 + 4.1 KV
good Q6_K 6.56 106.3 GB
102.2 + 4.1 KV
excellent Q8_0 8.5 136.4 GB
132.2 + 4.1 KV
lossless FP16 16 252.6 GB
248.5 + 4.1 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 PIXTRAL LARGE 124B 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 — 37.4 GB VRAM IQ2_M — 45.9 GB VRAM Q2_K — 49.5 GB VRAM IQ3_XXS — 50.9 GB VRAM IQ3_XS — 54.7 GB VRAM Q3_K_S — 56.9 GB VRAM IQ3_M — 58.8 GB VRAM Q3_K_M — 62.5 GB VRAM Q3_K_L — 67.1 GB VRAM IQ4_XS — 69.6 GB VRAM Q4_K_S — 72.9 GB VRAM Q4_K_M — 76.3 GB VRAM Q5_K_S — 86.8 GB VRAM Q5_K_M — 88.8 GB VRAM Q6_K — 102.2 GB VRAM Q8_0 — 132.2 GB VRAM FP16 — 248.5 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:124b-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 Pixtral Large 124B
Build Hardware for Pixtral Large 124B Pixtral Large 124B — most capable Mistral vision model. Strong reasoning with images.
Read full model card ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
Pixtral Large 124B — 124B Dense. ▸ SPECIFICATIONS
PARAMETERS 124B
ARCHITECTURE Dense Transformer
CONTEXT LENGTH 128K tokens
CAPABILITIES chat, vision, reasoning
RELEASE DATE 2024-11-18
PROVIDER Mistral AI
FAMILY mistral ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_XXS 2.38 37.4 GB 65% IQ2_M 2.93 45.9 GB 75% Q2_K 3.16 49.5 GB 78% IQ3_XXS 3.25 50.9 GB 82% IQ3_XS 3.5 54.7 GB 84% Q3_K_S 3.64 56.9 GB 85% IQ3_M 3.76 58.8 GB 86% Q3_K_M 4 62.5 GB 88% Q3_K_L 4.3 67.1 GB 90% IQ4_XS 4.46 69.6 GB 92% Q4_K_S 4.67 72.9 GB 93% Q4_K_M 4.89 76.3 GB 94% Q5_K_S 5.57 86.8 GB 96% Q5_K_M 5.7 88.8 GB 96% Q6_K 6.56 102.2 GB 97% Q8_0 8.5 132.2 GB 100% FP16 16 248.5 GB 100%
§ 01 BENCHMARK SCORES
HumanEval 82.0
MMLU-PRO 50.7
MATH 49.5
IFEval 87.0
BBH 52.7
MMMU 62.7
GPQA 24.9
MUSR 17.2
MMBench 79.0
GPQA Diamond 50.5
LiveCodeBench 26.1
AIME 2.3
MATH-500 71.4
HLE 3.6
AA Intelligence 14.0
AA Math 2.3
aa_ifbench 34.5
aa_tau2 36.5
aa_scicode 29.2
aa_lcr 10.3
§ 03 COMPATIBLE GPUs
30 @ Q4_K_M Feedback 