FITMYLLM · AUGUST 17, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
Quantization Options Select your GPU for speed estimates Context length: 4K 8K 16K 32K 64K 128K 250K
Quant Bits VRAM @ 16K Quality IQ2_XXS 2.38 36.5 GB
33.5 + 3.0 KV
low IQ2_M 2.93 44.1 GB
41.1 + 3.0 KV
low Q2_K 3.16 47.3 GB
44.3 + 3.0 KV
low IQ3_XXS 3.25 48.6 GB
45.6 + 3.0 KV
low IQ3_XS 3.5 52.1 GB
49.1 + 3.0 KV
low Q3_K_S 3.64 54.0 GB
51.0 + 3.0 KV
low IQ3_M 3.76 55.7 GB
52.7 + 3.0 KV
low Q3_K_M 4 59.0 GB
56.0 + 3.0 KV
low Q3_K_L 4.3 63.2 GB
60.2 + 3.0 KV
moderate IQ4_XS 4.46 65.4 GB
62.4 + 3.0 KV
moderate Q4_K_S 4.67 68.3 GB
65.3 + 3.0 KV
moderate Q4_K_M 4.89 71.3 GB
68.3 + 3.0 KV
good Q5_K_S 5.57 80.8 GB
77.8 + 3.0 KV
good Q5_K_M 5.7 82.6 GB
79.6 + 3.0 KV
good Q6_K 6.56 94.5 GB
91.5 + 3.0 KV
excellent Q8_0 8.5 121.4 GB
118.4 + 3.0 KV
lossless FP16 16 225.5 GB
222.5 + 3.0 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 COMMAND A 111B 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 — 33.5 GB VRAM IQ2_M — 41.1 GB VRAM Q2_K — 44.3 GB VRAM IQ3_XXS — 45.6 GB VRAM IQ3_XS — 49.1 GB VRAM Q3_K_S — 51.0 GB VRAM IQ3_M — 52.7 GB VRAM Q3_K_M — 56.0 GB VRAM Q3_K_L — 60.2 GB VRAM IQ4_XS — 62.4 GB VRAM Q4_K_S — 65.3 GB VRAM Q4_K_M — 68.3 GB VRAM Q5_K_S — 77.8 GB VRAM Q5_K_M — 79.6 GB VRAM Q6_K — 91.5 GB VRAM Q8_0 — 118.4 GB VRAM FP16 — 222.5 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 command-a:111b-03-2025-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 Command A 111B
Build Hardware for Command A 111B Command A — Cohere's most powerful model. 23 language support, strong agentic capabilities.
Read full model card ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
Command A 111B — 111B Dense. ▸ SPECIFICATIONS
PARAMETERS 111B
ARCHITECTURE Dense Transformer
CONTEXT LENGTH 250K tokens
CAPABILITIES chat, coding, reasoning, multilingual
RELEASE DATE 2025-03-01
PROVIDER Cohere
FAMILY command ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_XXS 2.38 33.5 GB 65% IQ2_M 2.93 41.1 GB 75% Q2_K 3.16 44.3 GB 78% IQ3_XXS 3.25 45.6 GB 82% IQ3_XS 3.5 49.1 GB 84% Q3_K_S 3.64 51.0 GB 85% IQ3_M 3.76 52.7 GB 86% Q3_K_M 4 56.0 GB 88% Q3_K_L 4.3 60.2 GB 90% IQ4_XS 4.46 62.4 GB 92% Q4_K_S 4.67 65.3 GB 93% Q4_K_M 4.89 68.3 GB 94% Q5_K_S 5.57 77.8 GB 96% Q5_K_M 5.7 79.6 GB 96% Q6_K 6.56 91.5 GB 97% Q8_0 8.5 118.4 GB 100% FP16 16 222.5 GB 100%
§ 01 BENCHMARK SCORES
MMLU-PRO 38.0
MATH 12.4
IFEval 75.4
BBH 42.8
GPQA 13.4
MUSR 19.8
BigCodeBench 33.8
Arena Elo 1481.0
GPQA Diamond 52.7
LiveCodeBench 28.7
AIME 13.0
MATH-500 81.9
HLE 4.6
AA Intelligence 13.5
AA Coding 9.9
AA Math 13.0
aa_ifbench 36.5
aa_terminal_bench 0.8
aa_tau2 15.2
aa_scicode 28.1
aa_lcr 18.0
§ 02 RUN COMMAND
Run Command A 111B locally with Ollama — needs 68.3 GB VRAM at Q4_K_M:
$ ollama run command-a:111b
§ 03 COMPATIBLE GPUs
30 @ Q4_K_M Feedback 