FITMYLLM · SEPTEMBER 6, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
355B
Parameters (12B active)
Quantization Options Select your GPU for speed estimates Context length: 4K 8K 16K 32K 64K 125K
Quant Bits VRAM @ 16K Quality IQ2_XXS 2.38 110.4 GB
106.1 + 4.3 KV
low IQ2_M 2.93 134.8 GB
130.5 + 4.3 KV
low Q2_K 3.16 145.0 GB
140.7 + 4.3 KV
low IQ3_XXS 3.25 149.0 GB
144.7 + 4.3 KV
low IQ3_XS 3.5 160.1 GB
155.8 + 4.3 KV
low Q3_K_S 3.64 166.3 GB
162.0 + 4.3 KV
low IQ3_M 3.76 171.7 GB
167.3 + 4.3 KV
low Q3_K_M 4 182.3 GB
178.0 + 4.3 KV
low Q3_K_L 4.3 195.6 GB
191.3 + 4.3 KV
moderate IQ4_XS 4.46 202.7 GB
198.4 + 4.3 KV
moderate Q4_K_S 4.67 212.0 GB
207.7 + 4.3 KV
moderate Q4_K_M 4.89 221.8 GB
217.5 + 4.3 KV
good Q5_K_S 5.57 252.0 GB
247.7 + 4.3 KV
good Q5_K_M 5.7 257.7 GB
253.4 + 4.3 KV
good Q6_K 6.56 295.9 GB
291.6 + 4.3 KV
excellent Q8_0 8.5 382.0 GB
377.7 + 4.3 KV
lossless FP16 16 714.8 GB
710.5 + 4.3 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 ~110 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 GLM 4.5 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 — 106.1 GB VRAM IQ2_M — 130.5 GB VRAM Q2_K — 140.7 GB VRAM IQ3_XXS — 144.7 GB VRAM IQ3_XS — 155.8 GB VRAM Q3_K_S — 162.0 GB VRAM IQ3_M — 167.3 GB VRAM Q3_K_M — 178.0 GB VRAM Q3_K_L — 191.3 GB VRAM IQ4_XS — 198.4 GB VRAM Q4_K_S — 207.7 GB VRAM Q4_K_M — 217.5 GB VRAM Q5_K_S — 247.7 GB VRAM Q5_K_M — 253.4 GB VRAM Q6_K — 291.6 GB VRAM Q8_0 — 377.7 GB VRAM FP16 — 710.5 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 glm:355b-q4_K_MCOPY
Tag may need adjustment — check ollama.com/library/glm 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 GLM 4.5
Build Hardware for GLM 4.5 GLM 4.5 — Tsinghua/Zhipu's flagship MoE model. Competitive with top proprietary models.
Read full model card ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
GLM 4.5 — 355B MoE. ▸ SPECIFICATIONS
PARAMETERS 355B (12B active)
ARCHITECTURE Mixture of Experts
CONTEXT LENGTH 125K tokens
CAPABILITIES chat, coding, reasoning, multilingual
RELEASE DATE 2025-08-08
PROVIDER Zhipu AI
FAMILY glm ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_XXS 2.38 106.1 GB 65% IQ2_M 2.93 130.5 GB 75% Q2_K 3.16 140.7 GB 78% IQ3_XXS 3.25 144.7 GB 82% IQ3_XS 3.5 155.8 GB 84% Q3_K_S 3.64 162.0 GB 85% IQ3_M 3.76 167.3 GB 86% Q3_K_M 4 178.0 GB 88% Q3_K_L 4.3 191.3 GB 90% IQ4_XS 4.46 198.4 GB 92% Q4_K_S 4.67 207.7 GB 93% Q4_K_M 4.89 217.5 GB 94% Q5_K_S 5.57 247.7 GB 96% Q5_K_M 5.7 253.4 GB 96% Q6_K 6.56 291.6 GB 97% Q8_0 8.5 377.7 GB 100% FP16 16 710.5 GB 100%
§ 01 BENCHMARK SCORES
MMLU-PRO 75.1
GPQA Diamond 78.2
LiveCodeBench 73.8
AIME 73.7
MATH-500 97.9
HLE 12.2
AA Intelligence 26.4
AA Coding 26.3
AA Math 73.7
aa_ifbench 34.2
aa_terminal_bench 5.3
aa_tau2 22.5
aa_scicode 22.1
aa_lcr 43.7
§ 03 COMPATIBLE GPUs
5 @ Q4_K_M Feedback 