FITMYLLM · AUGUST 17, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
480B
Parameters (35B active)
Quantization Options Select your GPU for speed estimates Context length: 4K 8K 16K 32K 64K 128K 256K
Quant Bits VRAM @ 16K Quality IQ2_XXS 2.38 146.2 GB
143.3 + 2.9 KV
low IQ2_M 2.93 179.2 GB
176.3 + 2.9 KV
low Q2_K 3.16 193.0 GB
190.1 + 2.9 KV
low IQ3_XXS 3.25 198.4 GB
195.5 + 2.9 KV
low IQ3_XS 3.5 213.4 GB
210.5 + 2.9 KV
low Q3_K_S 3.64 221.8 GB
218.9 + 2.9 KV
low IQ3_M 3.76 229.0 GB
226.1 + 2.9 KV
low Q3_K_M 4 243.4 GB
240.5 + 2.9 KV
low Q3_K_L 4.3 261.4 GB
258.5 + 2.9 KV
moderate IQ4_XS 4.46 271.0 GB
268.1 + 2.9 KV
moderate Q4_K_S 4.67 283.6 GB
280.7 + 2.9 KV
moderate Q4_K_M 4.89 296.8 GB
293.9 + 2.9 KV
good Q5_K_S 5.57 337.6 GB
334.7 + 2.9 KV
good Q5_K_M 5.7 345.4 GB
342.5 + 2.9 KV
good Q6_K 6.56 397.0 GB
394.1 + 2.9 KV
excellent Q8_0 8.5 513.4 GB
510.5 + 2.9 KV
lossless FP16 16 963.4 GB
960.5 + 2.9 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 ~146 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 QWEN3-CODER 480B-A35B 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 — 143.3 GB VRAM IQ2_M — 176.3 GB VRAM Q2_K — 190.1 GB VRAM IQ3_XXS — 195.5 GB VRAM IQ3_XS — 210.5 GB VRAM Q3_K_S — 218.9 GB VRAM IQ3_M — 226.1 GB VRAM Q3_K_M — 240.5 GB VRAM Q3_K_L — 258.5 GB VRAM IQ4_XS — 268.1 GB VRAM Q4_K_S — 280.7 GB VRAM Q4_K_M — 293.9 GB VRAM Q5_K_S — 334.7 GB VRAM Q5_K_M — 342.5 GB VRAM Q6_K — 394.1 GB VRAM Q8_0 — 510.5 GB VRAM FP16 — 960.5 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 qwen3-coder:480b-a35b-q4_K_MCOPY
Downloads and runs automatically. Add --verbose for speed stats.
▸ 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 Qwen3-Coder 480B-A35B
Build Hardware for Qwen3-Coder 480B-A35B ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
Qwen3-Coder 480B-A35B — 480B MoE. ▸ SPECIFICATIONS
PARAMETERS 480B (35B active)
ARCHITECTURE Mixture of Experts
CONTEXT LENGTH 256K tokens
CAPABILITIES coding, tool_use, reasoning, chat
RELEASE DATE 2025-11-01
PROVIDER Alibaba
FAMILY qwen ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_XXS 2.38 143.3 GB 65% IQ2_M 2.93 176.3 GB 75% Q2_K 3.16 190.1 GB 78% IQ3_XXS 3.25 195.5 GB 82% IQ3_XS 3.5 210.5 GB 84% Q3_K_S 3.64 218.9 GB 85% IQ3_M 3.76 226.1 GB 86% Q3_K_M 4 240.5 GB 88% Q3_K_L 4.3 258.5 GB 90% IQ4_XS 4.46 268.1 GB 92% Q4_K_S 4.67 280.7 GB 93% Q4_K_M 4.89 293.9 GB 94% Q5_K_S 5.57 334.7 GB 96% Q5_K_M 5.7 342.5 GB 96% Q6_K 6.56 394.1 GB 97% Q8_0 8.5 510.5 GB 100% FP16 16 960.5 GB 100%
§ 01 BENCHMARK SCORES
MMLU-PRO 78.8
LiveCodeBench 58.5
SWE-bench 69.6
AIME 39.3
MATH-500 94.2
GPQA Diamond 61.8
HLE 4.4
AA Intelligence 24.8
AA Coding 24.6
AA Math 39.3
aa_ifbench 40.5
aa_terminal_bench 18.9
aa_tau2 43.6
aa_scicode 35.9
aa_lcr 42.3
§ 02 RUN COMMAND
Run Qwen3-Coder 480B-A35B locally with Ollama — needs 293.9 GB VRAM at Q4_K_M:
$ ollama run qwen3-coder:480b
§ 03 COMPATIBLE GPUs
2 @ Q4_K_M Feedback 