FITMYLLM · AUGUST 17, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
132B
Parameters (36B active)
Quantization Options Select your GPU for speed estimates Context length: 4K 8K 16K 32K
Quant Bits VRAM @ 16K Quality IQ2_XXS 2.38 41.6 GB
39.8 + 1.9 KV
low IQ2_M 2.93 50.7 GB
48.8 + 1.9 KV
low Q2_K 3.16 54.5 GB
52.6 + 1.9 KV
low IQ3_XXS 3.25 56.0 GB
54.1 + 1.9 KV
low IQ3_XS 3.5 60.1 GB
58.2 + 1.9 KV
low Q3_K_S 3.64 62.4 GB
60.5 + 1.9 KV
low IQ3_M 3.76 64.4 GB
62.5 + 1.9 KV
low Q3_K_M 4 68.4 GB
66.5 + 1.9 KV
low Q3_K_L 4.3 73.3 GB
71.4 + 1.9 KV
moderate IQ4_XS 4.46 76.0 GB
74.1 + 1.9 KV
moderate Q4_K_S 4.67 79.4 GB
77.5 + 1.9 KV
moderate Q4_K_M 4.89 83.0 GB
81.2 + 1.9 KV
good Q5_K_S 5.57 94.3 GB
92.4 + 1.9 KV
good Q5_K_M 5.7 96.4 GB
94.5 + 1.9 KV
good Q6_K 6.56 110.6 GB
108.7 + 1.9 KV
excellent Q8_0 8.5 142.6 GB
140.7 + 1.9 KV
lossless FP16 16 266.4 GB
264.5 + 1.9 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 DBRX 132B 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 — 39.8 GB VRAM IQ2_M — 48.8 GB VRAM Q2_K — 52.6 GB VRAM IQ3_XXS — 54.1 GB VRAM IQ3_XS — 58.2 GB VRAM Q3_K_S — 60.5 GB VRAM IQ3_M — 62.5 GB VRAM Q3_K_M — 66.5 GB VRAM Q3_K_L — 71.4 GB VRAM IQ4_XS — 74.1 GB VRAM Q4_K_S — 77.5 GB VRAM Q4_K_M — 81.2 GB VRAM Q5_K_S — 92.4 GB VRAM Q5_K_M — 94.5 GB VRAM Q6_K — 108.7 GB VRAM Q8_0 — 140.7 GB VRAM FP16 — 264.5 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 dbrx:132bCOPY
Tag may need adjustment — check ollama.com/library/dbrx 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 DBRX 132B
Build Hardware for DBRX 132B Databricks DBRX — 132B MoE with 36B active params. Strong coding and general reasoning.
Read full model card ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
DBRX 132B — 132B MoE. ▸ SPECIFICATIONS
PARAMETERS 132B (36B active)
ARCHITECTURE Mixture of Experts
CONTEXT LENGTH 32K tokens
CAPABILITIES chat, coding
RELEASE DATE 2024-03-27
PROVIDER Databricks
FAMILY dbrx ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_XXS 2.38 39.8 GB 65% IQ2_M 2.93 48.8 GB 75% Q2_K 3.16 52.6 GB 78% IQ3_XXS 3.25 54.1 GB 82% IQ3_XS 3.5 58.2 GB 84% Q3_K_S 3.64 60.5 GB 85% IQ3_M 3.76 62.5 GB 86% Q3_K_M 4 66.5 GB 88% Q3_K_L 4.3 71.4 GB 90% IQ4_XS 4.46 74.1 GB 92% Q4_K_S 4.67 77.5 GB 93% Q4_K_M 4.89 81.2 GB 94% Q5_K_S 5.57 92.4 GB 96% Q5_K_M 5.7 94.5 GB 96% Q6_K 6.56 108.7 GB 97% Q8_0 8.5 140.7 GB 100% FP16 16 264.5 GB 100%
§ 01 BENCHMARK SCORES
HumanEval 70.1
MMLU-PRO 45.0
MATH 66.9
IFEval 62.0
BBH 56.0
GPQA 12.2
MUSR 12.2
§ 02 RUN COMMAND
Run DBRX 132B locally with Ollama — needs 81.2 GB VRAM at Q4_K_M:
$ ollama run dbrx:132b
§ 03 COMPATIBLE GPUs
30 @ Q4_K_M Feedback