Command R 35B
Cohere · CC-BY-NC-4.0
Cohere's RAG-optimized model with long-context reasoning support.
- Parameters
- 35B
- Architecture
- Dense
- Context
- 128,000 tokens
- Released
- 2024-03-11
- Engines
- llama.cpp, ollama, vLLM, TGI
Parameters
35B
VRAM
30 GB
Context
125K
Formats
6
GPUs
16
Command R 35B (35B) requires 30 GB VRAM at recommended quality (Q6_K). At efficient quality (Q4_K_M), it fits in 20 GB VRAM, making it compatible with the AMD Radeon RX 7900 XTX. On NVIDIA Grace Blackwell Ultra GB300, expect approximately 110 tok/s at Q8_0. For the best experience, High-End Home AI Server ($3,842) is recommended.
Source: OwnRig methodology
30 GB
Q6_K
26 GB
125K tokens
Chat
VRAM Requirements
| Quality | Quantization | VRAM | File Size |
|---|---|---|---|
| full | Q8_0 | 38.5 GB | 35 GB |
| recommended | Q6_K | 30 GB | 26 GB |
| recommended | Q5_K_M | 25 GB | 21.5 GB |
| efficient | Q4_K_M | 20 GB | 17.5 GB |
| compressed | Q3_K_M | 16 GB | 13.5 GB |
| compressed | Q2_K | 12.5 GB | 10.7 GB |
Context Length Impact
KV cache VRAM at Q6_K quality. Longer context = more memory.
| Context | KV Cache | Total VRAM |
|---|---|---|
| 2K | 614 MB | 30.6 GBexceeds 24 GB |
| 4K | 1.1 GB | 31.1 GBexceeds 24 GB |
| 8K | 2.2 GB | 32.2 GBexceeds 24 GB |
| 16K | 4.5 GB | 34.5 GBexceeds 24 GB |
| 32K | 9 GB | 39 GBexceeds 24 GB |
| 64K | 17.9 GB | 47.9 GBexceeds 24 GB |
Compatible GPUs
16 devices| NVIDIA Grace Blackwell Ultra GB300 | Q8_0 | 110 tok/s | Excellent |
| AMD Radeon RX 7900 XTX | Q4_K_M | – | Acceptable |
| AMD Radeon Pro W7900 | Q8_0 | – | Acceptable |
| NVIDIA RTX PRO 6000 Blackwell | Q8_0 | 28 tok/s | Acceptable |
| NVIDIA RTX PRO 6000 Blackwell Max-Q | Q8_0 | 26 tok/s | Acceptable |
| AMD Radeon RX 7600 | Q3_K_M | 2 tok/s | Marginal |
| AMD Radeon RX 9070 | Q3_K_M | 4 tok/s | Marginal |
| AMD Radeon RX 9060 XT 16GB | Q3_K_M | 2 tok/s | Marginal |
| Apple M3 Pro (18GB Unified) | Q3_K_M | – | Not viable |
| NVIDIA GeForce RTX 3080 10GB | Q2_K | – | Not viable |
| NVIDIA GeForce RTX 4060 8GB | Q2_K | – | Not viable |
| NVIDIA RTX 4060 Laptop (40-60W) | Q2_K | – | Not viable |
| NVIDIA RTX 4070 Laptop (80-115W) | Q2_K | – | Not viable |
| NVIDIA GeForce RTX 4070 Ti 12GB | Q2_K | – | Not viable |
| AMD Radeon RX 9060 XT 8GB | Q3_K_M | – | Not viable |
| NVIDIA GeForce RTX 5060 8GB | Q2_K | – | Not viable |
Showing 16 of 16 entries
Frequently Asked Questions
- How much VRAM does Command R 35B need?
- Command R 35B requires 30 GB VRAM at recommended quality (Q6_K). At lower quality settings, it can fit in as little as 12.5 GB.
- What is the best GPU for Command R 35B?
- The NVIDIA Grace Blackwell Ultra GB300 delivers the best performance for Command R 35B, achieving 110 tok/s at Q8_0 with an excellent rating.
- What quantization should I use for Command R 35B?
- For the best quality, use Q6_K (30 GB VRAM). If your GPU has limited VRAM, Q2_K (12.5 GB) is the most efficient option with acceptable quality.
Data confidence: estimated. Source
VRAM requirements are calculated from model parameters and may vary by inference engine, context length, and batch size. Performance estimates are based on community benchmarks and should be verified for your specific configuration.Cohere is a trademark of its respective owner. OwnRig is not affiliated with or endorsed by the model creator.