Plenty of companies are introducing coding agents these days. Mistral is no different. Devstral is its agentic LLM for software engineering tasks. It is designed to solve GitHub issues. As the company explains:
Devstral achieves a score of 46.8% on SWE-Bench Verified, outperforming prior open-source SoTA models by more than 6% points. When evaluated under the same test scaffold (OpenHands), Devstral exceeds far larger models such as Deepseek-V3-0324 (671B) and Qwen3 232B-A22B.
As the company suggests, Devstral surpasses GPT 4.1 mini by over 20%. It runs on RTX 4090. If you run it on a MAc, you are going to need 32GB RAM. Here is the pricing: ($0.1/M input tokens and $0.3/M output tokens).
[HT]
