FITMYLLM · AUGUST 17, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
400B
Parameters (17B active)
Quantization Options Select your GPU for speed estimates Context length: 4K 8K 16K 32K 64K 128K 256K 512K 1M
Quant Bits VRAM @ 16K Quality IQ2_XXS 2.38 121.7 GB
119.5 + 2.3 KV
low IQ2_M 2.93 149.2 GB
147.0 + 2.3 KV
low Q2_K 3.16 160.7 GB
158.5 + 2.3 KV
low IQ3_XXS 3.25 165.2 GB
163.0 + 2.3 KV
low IQ3_XS 3.5 177.7 GB
175.5 + 2.3 KV
low Q3_K_S 3.64 184.7 GB
182.5 + 2.3 KV
low IQ3_M 3.76 190.7 GB
188.5 + 2.3 KV
low Q3_K_M 4 202.7 GB
200.5 + 2.3 KV
low Q3_K_L 4.3 217.7 GB
215.5 + 2.3 KV
moderate IQ4_XS 4.46 225.7 GB
223.5 + 2.3 KV
moderate Q4_K_S 4.67 236.2 GB
234.0 + 2.3 KV
moderate Q4_K_M 4.89 247.2 GB
245.0 + 2.3 KV
good Q5_K_S 5.57 281.2 GB
279.0 + 2.3 KV
good Q5_K_M 5.7 287.7 GB
285.5 + 2.3 KV
good Q6_K 6.56 330.7 GB
328.5 + 2.3 KV
excellent Q8_0 8.5 427.7 GB
425.5 + 2.3 KV
lossless FP16 16 802.7 GB
800.5 + 2.3 KV
lossless
Select your GPU above to see speed estimates and compatibility for each quantization.
Too big for a single GPU — plan a multi-GPU deployment
Even the lightest quant needs ~122 GB. Size GPUs, replicas, TCO and scaling for a production setup. Open in Enterprise →
▸ READY TO RUN THIS? RENT BY THE HOUR
RENT A GPU AND RUN LLAMA-4-MAVERICK-17B-128E 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 — 119.5 GB VRAM IQ2_M — 147.0 GB VRAM Q2_K — 158.5 GB VRAM IQ3_XXS — 163.0 GB VRAM IQ3_XS — 175.5 GB VRAM Q3_K_S — 182.5 GB VRAM IQ3_M — 188.5 GB VRAM Q3_K_M — 200.5 GB VRAM Q3_K_L — 215.5 GB VRAM IQ4_XS — 223.5 GB VRAM Q4_K_S — 234.0 GB VRAM Q4_K_M — 245.0 GB VRAM Q5_K_S — 279.0 GB VRAM Q5_K_M — 285.5 GB VRAM Q6_K — 328.5 GB VRAM Q8_0 — 425.5 GB VRAM FP16 — 800.5 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 llama4:128x17bCOPY
Tag may need adjustment — check ollama.com/library/llama4 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-4-Maverick-17B-128E
Build Hardware for Llama-4-Maverick-17B-128E Llama 4 Maverick — Meta's 400B MoE. 128 experts, frontier-class performance.
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-4-Maverick-17B-128E — 400B MoE. ▸ SPECIFICATIONS
PARAMETERS 400B (17B active)
ARCHITECTURE Mixture of Experts
CONTEXT LENGTH 1024K tokens
CAPABILITIES chat, vision, reasoning
RELEASE DATE 2025-04-05
PROVIDER Meta
FAMILY llama ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_XXS 2.38 119.5 GB 65% IQ2_M 2.93 147.0 GB 75% Q2_K 3.16 158.5 GB 78% IQ3_XXS 3.25 163.0 GB 82% IQ3_XS 3.5 175.5 GB 84% Q3_K_S 3.64 182.5 GB 85% IQ3_M 3.76 188.5 GB 86% Q3_K_M 4 200.5 GB 88% Q3_K_L 4.3 215.5 GB 90% IQ4_XS 4.46 223.5 GB 92% Q4_K_S 4.67 234.0 GB 93% Q4_K_M 4.89 245.0 GB 94% Q5_K_S 5.57 279.0 GB 96% Q5_K_M 5.7 285.5 GB 96% Q6_K 6.56 328.5 GB 97% Q8_0 8.5 425.5 GB 100% FP16 16 800.5 GB 100%
§ 01 BENCHMARK SCORES
HumanEval 85.0
MMLU-PRO 69.0
MATH 78.0
IFEval 86.0
MMMU 61.0
MMBench 78.0
Arena Elo 1292.0
GPQA Diamond 67.1
LiveCodeBench 39.7
AIME 19.3
MATH-500 88.9
HLE 4.8
AA Intelligence 18.4
AA Coding 15.6
AA Math 19.3
aa_ifbench 43.0
aa_terminal_bench 6.8
aa_tau2 17.8
aa_scicode 33.1
aa_lcr 46.0
§ 02 RUN COMMAND
Run Llama-4-Maverick-17B-128E locally with Ollama — needs 245.0 GB VRAM at Q4_K_M:
$ ollama run llama4:128x17b
§ 03 COMPATIBLE GPUs
5 @ Q4_K_M Feedback 