FITMYLLM · NOVEMBER 4, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
314.84B
Parameters (13.2B 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 94.3 GB
94.2 + 0.2 KV
low IQ2_M 2.93 116.0 GB
115.8 + 0.2 KV
low Q2_K 3.16 125.0 GB
124.8 + 0.2 KV
low IQ3_XXS 3.25 128.6 GB
128.4 + 0.2 KV
low IQ3_XS 3.5 138.4 GB
138.2 + 0.2 KV
low Q3_K_S 3.64 143.9 GB
143.7 + 0.2 KV
low IQ3_M 3.76 148.6 GB
148.5 + 0.2 KV
low Q3_K_M 4 158.1 GB
157.9 + 0.2 KV
low Q3_K_L 4.3 169.9 GB
169.7 + 0.2 KV
moderate IQ4_XS 4.46 176.2 GB
176.0 + 0.2 KV
moderate Q4_K_S 4.67 184.4 GB
184.3 + 0.2 KV
moderate Q4_K_M 4.89 193.1 GB
192.9 + 0.2 KV
good Q5_K_S 5.57 219.9 GB
219.7 + 0.2 KV
good Q5_K_M 5.7 225.0 GB
224.8 + 0.2 KV
good Q6_K 6.56 258.8 GB
258.7 + 0.2 KV
excellent Q8_0 8.5 335.2 GB
335.0 + 0.2 KV
lossless FP16 16 630.3 GB
630.2 + 0.2 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 ~94 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 MOTIF 3 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 — 94.2 GB VRAM IQ2_M — 115.8 GB VRAM Q2_K — 124.8 GB VRAM IQ3_XXS — 128.4 GB VRAM IQ3_XS — 138.2 GB VRAM Q3_K_S — 143.7 GB VRAM IQ3_M — 148.5 GB VRAM Q3_K_M — 157.9 GB VRAM Q3_K_L — 169.7 GB VRAM IQ4_XS — 176.0 GB VRAM Q4_K_S — 184.3 GB VRAM Q4_K_M — 192.9 GB VRAM Q5_K_S — 219.7 GB VRAM Q5_K_M — 224.8 GB VRAM Q6_K — 258.7 GB VRAM Q8_0 — 335.0 GB VRAM FP16 — 630.2 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 other:315b-q4_K_MCOPY
Tag may need adjustment — check ollama.com/library/other 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 Motif 3
Build Hardware for Motif 3 ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
Motif 3 — 314.84B MoE. ▸ SPECIFICATIONS
PARAMETERS 314.84B (13.2B active)
ARCHITECTURE Mixture of Experts
CONTEXT LENGTH 256K tokens
CAPABILITIES chat, coding, reasoning, multilingual, math, agentic, tool_use
RELEASE DATE 2026-08-10
PROVIDER Motif Technologies
FAMILY other ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_XXS 2.38 94.2 GB 65% IQ2_M 2.93 115.8 GB 75% Q2_K 3.16 124.8 GB 78% IQ3_XXS 3.25 128.4 GB 82% IQ3_XS 3.5 138.2 GB 84% Q3_K_S 3.64 143.7 GB 85% IQ3_M 3.76 148.5 GB 86% Q3_K_M 4 157.9 GB 88% Q3_K_L 4.3 169.7 GB 90% IQ4_XS 4.46 176.0 GB 92% Q4_K_S 4.67 184.3 GB 93% Q4_K_M 4.89 192.9 GB 94% Q5_K_S 5.57 219.7 GB 96% Q5_K_M 5.7 224.8 GB 96% Q6_K 6.56 258.7 GB 97% Q8_0 8.5 335.0 GB 100% FP16 16 630.2 GB 100%
§ 01 BENCHMARK SCORES
GPQA Diamond 83.4
HLE 37.0
aa_scicode 42.2
aa_lcr 76.0
SWE-bench 76.2
aa_terminal_bench_v2 74.9
§ 03 COMPATIBLE GPUs
5 @ Q4_K_M Feedback