back to local model checker

Command R 35B

Tools

Cohere · 35B · Dense

Optimized for retrieval-augmented generation

CC BY-NC 4.0
Downloads

12.1K

Likes

1.1K

Released

2024-03

Context

128K tokens

Use cases

chatrag

Your hardware

16 GB Reference Desktop

Quant
Q4_K_M
Memory
18.4 GB
Speed
~6 tok/s
Fit
Barely runs

Quantization options

What is quantization?
QuantBitsVRAMDiskQualityStatus
Q2_K211.7 GB10.7 GBlowTight fit · ~17 t/s
Q3_K_M316.2 GB15 GBmoderateBarely runs · ~10 t/s
Q4_K_M418.4 GB17.1 GBgoodBarely runs · ~6 t/s
Q5_K_M522.9 GB21.4 GBgoodBarely runs · ~3 t/s
Q6_K627.4 GB25.7 GBexcellentBarely runs · ~2 t/s
Q8_0836.4 GB34.2 GBexcellentBarely runs · ~1 t/s
F161672.2 GB68.5 GBlosslessToo heavy · ~1 t/s

About this model

Command R 35B is a 35B dense transformer from Cohere in the Command family, designed primarily for chat, rag workflows. Supports structured tool calling; use Q4_K_M or higher when strict JSON schema adherence matters. Supports up to 128K context tokens; start at 4K–8K with Q8_0 KV cache on tight GPUs.

Hardware planning baselines start at 19.6 GB minimum memory and 32.6 GB recommended memory. At Q4_K_M, weights plus baseline runtime overhead use roughly 18.4 GB of RAM/VRAM and 17.1 GB of disk space. At 35B, Command R 35B belongs firmly in the workstation and multi-GPU tier. Running it comfortably requires 32.6 GB of total memory across dual RTX GPUs, high-memory Apple Silicon (Mac Studio Ultra/Max), or MoE hybrid CPU-RAM offloading—mobile devices and standard 16 GB laptops are too constrained for practical speeds.

Deploy Command R 35B locally via Ollama (`command-r:35b`), Ollama with Flash Attention enabled. Under Creative Commons Attribution-NonCommercial 4.0 International, review official weight cards before commercial redistribution.

Highlights

  • Local Fit Baseline19.6 GB minimum and 32.6 GB recommended RAM/VRAM.
  • Primary QuantQ4_K_M targets 18.4 GB memory and 17.1 GB disk.
  • Context & Capabilities128K max context · Dense · Tool calling.

System requirements by device

Memory planning targets 19.6 GB minimum RAM/VRAM and 32.6 GB recommended memory. The baseline quant is Q4_K_M, estimated at 18.4 GB memory and 17.1 GB disk. Actual runtime usage shifts with context length, KV cache precision, and OS background overhead.

DeviceNotes
Mac (Apple silicon)Requires a 48 GB–192 GB Mac Studio or MacBook Pro Max/Ultra with enough unified wired budget for 18.4 GB.
Windows / Linux GPUUse dual/multi-GPU CUDA splitting (--tensor-split) or MoE CPU expert offloading (-ot) paired with 32–128 GB DDR5 system RAM.
Phones & Compact Edge BoardsNot practical on phones (19.6 GB min RAM exceeds Android LMK and iOS Jetsam limits). Pick a 2B–4B model on mobile.

Start with a short session and leave memory for the operating system. A file fitting on disk does not guarantee that it will load in memory.

Estimated memory by quantization

These are rough estimates, not tested minimum requirements. Check the actual file and runtime documentation; context, offloading and operating-system overhead can change the memory needed.

Quant TierEstimated memoryDisk SizeQuality
Q2_K11.7 GB10.7 GBlow
Q3_K_M16.2 GB15.0 GBmoderate
Q4_K_M18.4 GB17.1 GBgood
Q5_K_M22.9 GB21.4 GBgood
Q6_K27.4 GB25.7 GBexcellent
Q8_036.4 GB34.2 GBexcellent
F1672.2 GB68.5 GBlossless

How to run Command R 35B on your device

Start with a short prompt and the smallest quant that still gives the quality you need. A clean first run is more useful than loading a larger file with no memory headroom.

Windows

Run Command R 35B on Multi-GPU or High-RAM Desktop

  1. Plan for 19.6 GB minimum combined memory and 32.6 GB recommended memory for Q4_K_M.
  2. Use `llama-server --split-mode layer --tensor-split` across multiple NVIDIA GPUs or high-channel DDR5 server RAM.
  3. Enable Flash Attention (`-fa on`) and Q8_0 KV cache (`-ctk q8_0 -ctv q8_0`) to prevent context buffers from exceeding GPU headroom.
  4. Avoid attempting to run workstation-scale weights on smartphones or 8 GB laptops; use a 2B–8B checkpoint on portable devices.
Device how-to guide
Mac

Run Command R 35B on High-Memory Apple Silicon Mac

  1. Verify you are on a 48 GB–192 GB Mac Studio or MacBook Pro Max/Ultra capable of holding 18.4 GB in wired unified memory.
  2. Launch via `llama-server` or LM Studio with MLX/Metal enabled and set initial context conservatively to 4K–8K.
  3. Monitor Activity Monitor Memory Pressure to confirm zero SSD swap thrashing during token generation.
  4. Step down to a lighter quantization (e.g., Q3_K_M or IQ4_XS) if total memory utilization exceeds 75%.
Device how-to guide

Command R 35B vs alternatives

ModelParamsContextMin RAMLicense
Command R 35B35B128K19.6 GBCC BY-NC 4.0
Qwen 3.5 35B-A3B35B (3B active)256K19.6 GBApache 2.0
Agents-A1 35B-A3B35B (3B active)256K19.6 GBApache 2.0
Ornith 1.0 35B-A3B35B (3B active)256K19.6 GBMIT
Qwen 3.6 35B-A3B36B (3B active)256K20.1 GBApache 2.0
Gemma 4 31B IT33B256K18.4 GBGemma

Use this table to choose the smallest nearby model that still matches the job. A related model with lower RAM can be the better local choice when the current page fits only at the edge of available memory.

Frequently asked questions

What is Command R 35B?
Command R 35B is a 35B dense transformer from Cohere, supporting up to 128K tokens with 19.6 GB minimum RAM and 32.6 GB recommended RAM.
How much VRAM or RAM does Command R 35B require?
At Q4_K_M, Command R 35B occupies roughly 18.4 GB of active memory and 17.1 GB on disk. Add 2–6 GB of spare headroom for KV cache buffers, runtime scratch space, and OS apps (32.6 GB recommended).
What is the best way to run Command R 35B locally?
Use `ollama run command-r:35b` for a quick CLI/API setup, or load Q4_K_M in LM Studio or llama.cpp with Flash Attention enabled when you want explicit KV cache and layer control.
Can Command R 35B run on a consumer laptop or smartphone?
No smartphone or 8–16 GB laptop can hold Command R 35B cleanly in local memory. It targets multi-GPU NVIDIA workstations, high-RAM Mac Studios, or MoE CPU-RAM hybrid rigs with at least 19.6 GB total memory.
How does Command R 35B compare to Qwen 3.5 35B-A3B?
Qwen 3.5 35B-A3B (35B, 19.6 GB min RAM) sits closest in size. Compare active parameters, context ceiling (128K vs 256K), and license terms (CC BY-NC 4.0) before switching.