FITMYLLM · NOVEMBER 4, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
321.32B
Parameters (18B 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 96.2 GB
96.1 + 0.1 KV
low IQ2_M 2.93 118.3 GB
118.2 + 0.1 KV
low Q2_K 3.16 127.5 GB
127.4 + 0.1 KV
low IQ3_XXS 3.25 131.2 GB
131.0 + 0.1 KV
low IQ3_XS 3.5 141.2 GB
141.1 + 0.1 KV
low Q3_K_S 3.64 146.8 GB
146.7 + 0.1 KV
low IQ3_M 3.76 151.6 GB
151.5 + 0.1 KV
low Q3_K_M 4 161.3 GB
161.1 + 0.1 KV
low Q3_K_L 4.3 173.3 GB
173.2 + 0.1 KV
moderate IQ4_XS 4.46 179.8 GB
179.6 + 0.1 KV
moderate Q4_K_S 4.67 188.2 GB
188.1 + 0.1 KV
moderate Q4_K_M 4.89 197.0 GB
196.9 + 0.1 KV
good Q5_K_S 5.57 224.3 GB
224.2 + 0.1 KV
good Q5_K_M 5.7 229.6 GB
229.4 + 0.1 KV
good Q6_K 6.56 264.1 GB
264.0 + 0.1 KV
excellent Q8_0 8.5 342.0 GB
341.9 + 0.1 KV
lossless FP16 16 643.3 GB
643.1 + 0.1 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 ~96 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-5.3-FLASH 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 — 96.1 GB VRAM IQ2_M — 118.2 GB VRAM Q2_K — 127.4 GB VRAM IQ3_XXS — 131.0 GB VRAM IQ3_XS — 141.1 GB VRAM Q3_K_S — 146.7 GB VRAM IQ3_M — 151.5 GB VRAM Q3_K_M — 161.1 GB VRAM Q3_K_L — 173.2 GB VRAM IQ4_XS — 179.6 GB VRAM Q4_K_S — 188.1 GB VRAM Q4_K_M — 196.9 GB VRAM Q5_K_S — 224.2 GB VRAM Q5_K_M — 229.4 GB VRAM Q6_K — 264.0 GB VRAM Q8_0 — 341.9 GB VRAM FP16 — 643.1 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:321b-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-5.3-Flash
Build Hardware for GLM-5.3-Flash ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
GLM-5.3-Flash — 321.32B MoE. ▸ SPECIFICATIONS
PARAMETERS 321.32B (18B active)
ARCHITECTURE Mixture of Experts
CONTEXT LENGTH 1024K tokens
CAPABILITIES chat, coding, reasoning, multilingual, vision, math, agentic, tool_use
RELEASE DATE 2026-08-26
PROVIDER Zhipu AI
FAMILY glm ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_XXS 2.38 96.1 GB 65% IQ2_M 2.93 118.2 GB 75% Q2_K 3.16 127.4 GB 78% IQ3_XXS 3.25 131.0 GB 82% IQ3_XS 3.5 141.1 GB 84% Q3_K_S 3.64 146.7 GB 85% IQ3_M 3.76 151.5 GB 86% Q3_K_M 4 161.1 GB 88% Q3_K_L 4.3 173.2 GB 90% IQ4_XS 4.46 179.6 GB 92% Q4_K_S 4.67 188.1 GB 93% Q4_K_M 4.89 196.9 GB 94% Q5_K_S 5.57 224.2 GB 96% Q5_K_M 5.7 229.4 GB 96% Q6_K 6.56 264.0 GB 97% Q8_0 8.5 341.9 GB 100% FP16 16 643.1 GB 100%
§ 01 BENCHMARK SCORES
GPQA Diamond 91.2
HLE 39.9
aa_scicode 51.6
aa_lcr 80.0
aa_terminal_bench_v2 84.3
aa_terminal_bench_v4 32.8
§ 03 COMPATIBLE GPUs
5 @ Q4_K_M Feedback 