CohereLabs · text

c4ai-command-r-plus

CohereLabs/c4ai-command-r-plus

c4ai-command-r-plus at Q4_K_M is exactly 62,750,607,680 bytes (58.44 GiB / 62.75 GB) — an effective 4.836 bits per weight, not the nominal 4.

From the file· summed from 2 file(s)
Parameters
104B
Architecture
command-r
Context
—
native (config.json)
License
cc-by-nc-4.0

Shipped quantizations

● exact bytes, summed from published files
QuantSize●Exact bytes●Effective bpw●Tensors●Publisher
IQ1_S21.59 GiB23,181,871,5201.786—dranger003
IQ1_S21.59 GiB23,181,871,5201.786—qwp4w3hyb
IQ1_M23.49 GiB25,217,943,9681.943—dranger003
IQ2_XXS26.65 GiB28,611,398,0482.205—qwp4w3hyb
IQ2_XXS26.65 GiB28,611,398,0482.205—dranger003
IQ2_XS29.46 GiB31,628,151,2002.437—qwp4w3hyb
IQ2_XS29.46 GiB31,628,151,2002.437—dranger003
IQ2_S31.04 GiB33,324,485,0242.568—qwp4w3hyb
IQ2_S31.04 GiB33,324,485,0242.568—dranger003
IQ2_M33.56 GiB36,039,248,2882.777—dranger003
IQ2_M33.56 GiB36,039,248,2882.777—qwp4w3hyb
Q2_K_S34.08 GiB36,595,452,3202.820—dranger003
Q2_K36.78 GiB39,497,386,1123.044—pmysl
Q2_K36.78 GiB39,497,386,4003.044—dranger003
IQ3_XXS37.87 GiB40,658,749,8563.133—dranger003
IQ3_XXS37.87 GiB40,658,749,8563.133—qwp4w3hyb
IQ3_XS40.61 GiB43,599,415,7123.360—dranger003
IQ3_XS40.61 GiB43,599,415,7123.360—qwp4w3hyb
Q3_K_S42.70 GiB45,851,756,6723.534—pmysl
IQ3_S42.80 GiB45,958,711,7123.542—qwp4w3hyb
IQ3_S42.80 GiB45,958,711,7123.542—dranger003
IQ3_M44.41 GiB47,683,357,0883.675—qwp4w3hyb
IQ3_M44.41 GiB47,683,357,0883.675—dranger003
Q3_K_M2 shards47.48 GiB50,982,439,2323.929—pmysl
Q3_K_M2 shards47.48 GiB50,982,439,5523.929—dranger003
Q3_K_L2 shards51.60 GiB55,402,187,0724.269—pmysl
Q3_K_L2 shards51.60 GiB55,402,187,3924.269—dranger003
IQ4_XS2 shards52.34 GiB56,201,202,3044.331—dranger003
IQ4_NL2 shards55.25 GiB59,321,764,4804.572—dranger003
Q4_K_S2 shards55.55 GiB59,642,628,4164.596—pmysl
Q4_K_S2 shards55.55 GiB59,642,628,7044.596—dranger003
Q4_K_M2 shards58.44 GiB62,750,607,6804.836—pmysl
Q4_K_M2 shards58.44 GiB62,750,607,9684.836—dranger003
Q5_K_S2 shards66.87 GiB71,804,012,8645.534—pmysl
Q5_K_S2 shards66.87 GiB71,804,013,1845.534—dranger003
Q5_K_M2 shards68.57 GiB73,622,243,6485.674—pmysl
Q5_K_M2 shards68.57 GiB73,622,243,9685.674—dranger003
Q6_K2 shards79.32 GiB85,173,356,8646.564—pmysl
Q6_K2 shards79.32 GiB85,173,356,8646.564—dranger003
Q8_03 shards102.74 GiB110,314,604,9928.501—pmysl

Compare with

same modality, comparable size
vs Qwen3-Coder-30B-A3B-Instructvs Qwen3.6-27Bvs Qwen3.8-27Bvs gemma-4-E4B-itvs Qwen3-30B-A3B-Thinking-2507vs Qwen3-8BAll alternatives →Will it fit on my card? →

Will it run on your card?

full quant x context sweep
By VRAM:8GB12GB16GB24GB32GB48GB
Radeon RX 6500 XT 4GBGeForce RTX 3050 6GBGeForce RTX 5050 8GBGeForce RTX 3080 10GBGeForce RTX 2080 Ti 11GBGeForce RTX 5070 12GBGeForce RTX 5060 Ti 16GBApple M3 Pro 18GBGeForce RTX 3080 Ti 20GBGeForce RTX 5090 D V2 24GBGeForce RTX 5090 D 32GBApple M5 Max 36GBApple M5 Pro 48GBApple M5 Max 64GBApple M3 Ultra 96GBApple M5 Max 128GBApple M2 Ultra 192GBApple M3 Ultra 256GBApple M3 Ultra 512GB

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts Q4_K_M at roughly 54.38 GiB. The real file is 58.44 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

Architecture unavailable — this repository is gated and no ungated mirror was found. Exact file sizes above are still authoritative; only the KV math needs the config.

Questions people ask

How much VRAM does c4ai-command-r-plus need?
Q4_K_M is exactly 62,750,607,680 bytes (58.44 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
Which quantization of c4ai-command-r-plus should I use?
Q4_K_M is the usual default. Pick the largest quantization that fits your card at the context you actually need — the table above gives exact sizes for every one published.