Skip to content

Devstral 2 123B Instruct 2512

Mistral AI
Code Multilingual Tool Calls

Devstral 2 123B Instruct 2512 is a 125.03-billion-parameter dense transformer from Mistral AI, purpose-built for agentic software engineering and multi-file code editing. It achieves 72.2% on SWE-bench Verified, placing it among the strongest open-weight coding models available. The model supports tool calling and 11 languages including English, French, German, and Chinese. With a 256K context window and flash attention, it handles repository-level reasoning across large codebases and can be self-hosted with GGUF quantization.

Hardware Configuration

Optional — for precise deployment recommendations
Quantization Quality Size Fit
Q8_0 High 123.73 GB
Q8_K_XL High 136.03 GB
Q6_K High 95.53 GB
Q6_K_XL High 102.29 GB
Q5_K_M Medium 82.25 GB
Q5_K_S Medium 80.27 GB
Q5_K_XL Medium 82.19 GB
Q4_K_M Medium 69.76 GB
Q4_K_S Medium 66.36 GB
Q4_K_XL Medium 70.32 GB
Q4_0 Medium 66.12 GB
Q4_1 Medium 73.08 GB
Q3_K_M Low 56.45 GB
Q3_K_S Low 50.63 GB
Q3_K_XL Low 57.68 GB
Q2_K Low 43.39 GB
Q2_K_L Low 43.74 GB
Q2_K_XL Low 44.65 GB
Last updated: March 5, 2026