Codestral Mamba 7B

Overview

Mistral AI's pure Mamba SSM architecture for code, with linear-time inference and a 256k context window. Apache 2.0, but tooling support is still patchy.

When to pick this model

• Long-context code analysis across entire repositories
• Research into state-space models for code
• Inference workloads where constant memory matters more than raw quality
• Settings where mistral-inference or vLLM is already in the stack

VRAM requirements by quantization

Quantization	VRAM required
Q4_K_M (recommended)	5 GB
Q5_K_M	6 GB
Q8_0	9 GB
FP16 (no quantization)	14 GB

VRAM figures include model weights plus a typical 8k KV cache and ~600 MB runtime overhead (Ollama / llama.cpp baseline). Add headroom for higher context lengths.

Strengths

• Verified 256k context for whole-repo reasoning
• Constant memory footprint regardless of sequence length
• Apache 2.0 license
• Linear-time inference scales gracefully on long inputs

Limitations

• No official Ollama support
• Only partial llama.cpp integration
• Requires mistral-inference or vLLM for full functionality
• Quality trails transformer-based coders of similar size

Architecture & training

Architecture: Pure Mamba2 SSM · linear inference

Training: First serious Mamba for code.

Quick start

# HuggingFace : mistralai/Mamba-Codestral-7B-v0.1

Or use the open-source MCP server to query this model from Claude Desktop, Cursor, or any MCP-compatible client.