How much VRAM does Codestral 22B need?

Codestral 22B requires 15.1 GB VRAM at recommended quality (Q5_K_M). At lower quality settings, it can fit in as little as 10.3 GB.

What is the best GPU for Codestral 22B?

The NVIDIA Grace Blackwell Ultra GB300 delivers the best performance for Codestral 22B, achieving 180 tok/s at Q5_K_M with an excellent rating.

Can I run Codestral 22B on an RTX 4060 Ti?

Yes. On the NVIDIA GeForce RTX 4060 Ti 16GB, Codestral 22B runs at 18 tok/s (Q3_K_M, acceptable).

What quantization should I use for Codestral 22B?

For the best quality, use Q5_K_M (15.1 GB VRAM). If your GPU has limited VRAM, Q3_K_M (10.3 GB) is the most efficient option with acceptable quality.

Is Codestral 22B good for coding?

Yes. Codestral 22B is used with Cursor, Continue, Windsurf for local AI coding. For the best coding experience, pair it with an embedding model for local RAG.

CodingAI codingAI building22.2B

Coding

Codestral 22B

Mistral · Mistral AI Non-Production License

Mistral's dedicated coding model. Built for code completion and generation across 80+ languages. Fits on a single 16 GB GPU at Q3/Q4. Non-production license limits commercial use.

Parameters: 22.2B
Architecture: Dense
Context: 32,768 tokens
Released: 2024-05-29
Engines: llama.cpp, ollama, vLLM
Builder Tools: Cursor, Continue, Windsurf

Parameters

22.2B

VRAM

15.1 GB

Context

32K

Formats

GPUs

Codestral 22B (22.2B) requires 15.1 GB VRAM at recommended quality (Q5_K_M). At efficient quality (Q4_K_M), it fits in 12.7 GB VRAM, making it compatible with the NVIDIA RTX 4080 Laptop (120-150W). On NVIDIA Grace Blackwell Ultra GB300, expect approximately 180 tok/s at Q5_K_M. For the best experience, Budget Home AI Server ($1,162) is recommended.

Source: OwnRig methodology

VRAM (Recommended)

15.1 GB

Quantization

Q5_K_M

File Size

13.3 GB

Max Context

32K tokens

Primary Use

Coding

Memory

VRAM Requirements

Quality	Quantization	VRAM	File Size
recommended	Q5_K_M	15.1 GB	13.3 GB
efficient	Q4_K_M	12.7 GB	11.1 GB
compressed	Q3_K_M	10.3 GB	8.7 GB

Scaling

Context Length Impact

KV cache VRAM at Q5_K_M quality. Longer context = more memory.

Context	KV Cache	Total VRAM
2K	307 MB	15.4 GB
4K	512 MB	15.6 GB
8K	1 GB	16.1 GB
16K	2 GB	17.1 GB
32K	4.1 GB	19.2 GB

Compatible GPUs

21 devices


NVIDIA Grace Blackwell Ultra GB300	Q5_K_M	180 tok/s	Excellent
NVIDIA GeForce RTX 4090	Q5_K_M	35 tok/s	Excellent
AMD Radeon Pro W7900	Q5_K_M	38 tok/s	Excellent
AMD Radeon RX 7900 XTX	Q5_K_M	30 tok/s	Good
NVIDIA RTX PRO 6000 Blackwell	Q5_K_M	73 tok/s	Good
NVIDIA RTX PRO 6000 Blackwell Max-Q	Q5_K_M	67 tok/s	Good
AMD Radeon RX 9070	Q3_K_M	32 tok/s	Good
NVIDIA GeForce RTX 4060 Ti 16GB	Q3_K_M	18 tok/s	Acceptable
NVIDIA RTX 4090 Laptop (150-175W)	Q3_K_M	15 tok/s	Acceptable
AMD Radeon RX 9060 XT 16GB	Q3_K_M	16 tok/s	Acceptable
NVIDIA GeForce RTX 5060 Ti 16GB	Q3_K_M	20 tok/s	Acceptable
NVIDIA GeForce RTX 4070 Ti 12GB	Q3_K_M	12 tok/s	Marginal
NVIDIA RTX 4080 Laptop (120-150W)	Q3_K_M	8 tok/s	Marginal
AMD Radeon RX 7600	Q3_K_M	2 tok/s	Marginal
Apple M3 Pro (18GB Unified)	Q3_K_M	–	Not viable
NVIDIA GeForce RTX 3080 10GB	Q3_K_M	–	Not viable
NVIDIA GeForce RTX 4060 8GB	Q3_K_M	–	Not viable
NVIDIA RTX 4060 Laptop (40-60W)	Q3_K_M	–	Not viable
NVIDIA RTX 4070 Laptop (80-115W)	Q3_K_M	–	Not viable
AMD Radeon RX 9060 XT 8GB	Q3_K_M	–	Not viable
NVIDIA GeForce RTX 5060 8GB	Q3_K_M	–	Not viable

Showing 21 of 21 entries

Builder Context

Codestral 22B is commonly used with Cursor, Continue, Windsurf. For an AI coding workflow, pair it with an embedding model like nomic-embed-text for local RAG.

Hardware

Recommended Builds

Complete PC builds that can run Codestral 22B.

Mid-range

AI Builder Workstation

Run every AI tool you need. Nothing leaves your machine

RTX 4090·24 GBVRAM

Runs 10 models

$2,773

Mid-range

Mid-Range AI Workstation

The sweet spot for AI: handles most models without overspending

RTX 4060 Ti 16GB·16 GBVRAM

Runs 8 models

$1,119

FAQ

Frequently Asked Questions

How much VRAM does Codestral 22B need?: Codestral 22B requires 15.1 GB VRAM at recommended quality (Q5_K_M). At lower quality settings, it can fit in as little as 10.3 GB.
What is the best GPU for Codestral 22B?: The NVIDIA Grace Blackwell Ultra GB300 delivers the best performance for Codestral 22B, achieving 180 tok/s at Q5_K_M with an excellent rating.
Can I run Codestral 22B on an RTX 4060 Ti?: Yes. On the NVIDIA GeForce RTX 4060 Ti 16GB, Codestral 22B runs at 18 tok/s (Q3_K_M, acceptable).
What quantization should I use for Codestral 22B?: For the best quality, use Q5_K_M (15.1 GB VRAM). If your GPU has limited VRAM, Q3_K_M (10.3 GB) is the most efficient option with acceptable quality.
Is Codestral 22B good for coding?: Yes. Codestral 22B is used with Cursor, Continue, Windsurf for local AI coding. For the best coding experience, pair it with an embedding model for local RAG.

All models

Data confidence: verified. 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.Mistral is a trademark of its respective owner. OwnRig is not affiliated with or endorsed by the model creator.