French AI company Mistral AI released its open-source model Mistral Medium 3.5, but the market response has been lukewarm.

BlockBeats reported on May 1st that French AI startup Mistral AI released its next-generation open-source model, Mistral Medium 3.5, on April 29th. This model features a dense architecture with 128 billion parameters and simultaneously launched the remote coding agent tool Mistral Vibe CLI and Le Chat's multi-step AI agent functionality. Official data shows that Mistral Medium 3.5 scored 77.6% in the SWE-Bench Verified programming benchmark and supports unified inference and coding capabilities. However, its API pricing, at $1.5 per million input tokens and $7.5 per million output tokens, significantly higher than Chinese open-source models, has sparked widespread criticism from the community. Currently, the top open-source model rankings are dominated by Chinese models such as Alibaba's Qwen, Zhipu GLM, and Xiaomi's MiMo-V2. Among them, Qwen 3.6, with only 27 billion parameters, achieved 72.4% on SWE-Bench and is completely open-source and free. However, some developers believe that Mistral's greatest value currently lies in its "non-US, non-Chinese" European identity. Due to its compliance with GDPR and local European deployment requirements, its model has been used by large European institutions, including HSBC, for private deployments. Market observers believe that Mistral's competitiveness stems more from geopolitics and corporate compliance than from pure model performance.