Command A+ is an open source model with 25 billion active parameters and 218B total parameters model optimized for agentic, multilingual, and reasoning-heavy tasks with a focus on enterprise performance, while also providing support for vision inp...
chatcodingreasoningmultilingualagentictool_use
218B
Parameters (25B active)
195K
Context length
13
Benchmarks
17
Quantizations
0
Architecture
MoE
Released
2026-05-15
Layers
32
KV Heads
8
Head Dim
128
Family
command
Quantization Options
Context length:
Quant
Bits
VRAM @ 16K
Quality
IQ2_XXS
2.38
66.1 GB
65.3 + 0.8 KV
low
IQ2_M
2.93
81.1 GB
80.3 + 0.8 KV
low
Q2_K
3.16
87.3 GB
86.6 + 0.8 KV
low
IQ3_XXS
3.25
89.8 GB
89.1 + 0.8 KV
low
IQ3_XS
3.5
96.6 GB
95.9 + 0.8 KV
low
Q3_K_S
3.64
100.4 GB
99.7 + 0.8 KV
low
IQ3_M
3.76
103.7 GB
102.9 + 0.8 KV
low
Q3_K_M
4
110.2 GB
109.5 + 0.8 KV
low
Q3_K_L
4.3
118.4 GB
117.7 + 0.8 KV
moderate
IQ4_XS
4.46
122.8 GB
122.0 + 0.8 KV
moderate
Q4_K_S
4.67
128.5 GB
127.7 + 0.8 KV
moderate
Q4_K_M
4.89
134.5 GB
133.7 + 0.8 KV
good
Q5_K_S
5.57
153.0 GB
152.3 + 0.8 KV
good
Q5_K_M
5.7
156.6 GB
155.8 + 0.8 KV
good
Q6_K
6.56
180.0 GB
179.2 + 0.8 KV
excellent
Q8_0
8.5
232.9 GB
232.1 + 0.8 KV
lossless
FP16
16
437.2 GB
436.5 + 0.8 KV
lossless
Select your GPU above to see speed estimates and compatibility for each quantization.
Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.
