Mistral AI retires NeMo 12B model and API endpoint July 31, 2026

Mistral AI is retiring the Mistral NeMo 12B model and its associated API endpoint, open-mistral-nemo-2407, with deprecation on May 22, 2026, and full shutdown on July 31, 2026. This model, a collaboration with NVIDIA, was noted for its 128k context window, Apache 2.0 license, efficient Tekken tokenizer, and affordability, making it a popular choice for local and budget-conscious deployments. Its retirement, alongside other Mistral models, signals a shift towards more complex architectures like Mixture-of-Experts and potentially higher costs for similar context lengths. AI

IMPACT Signals a shift in Mistral's model strategy towards more complex architectures and potentially higher costs for large context windows.

RANK_REASON The retirement of a specific model and its API endpoint by a major AI provider like Mistral AI, impacting users and requiring migration. [lever_c_demoted from significant: ic=1 ai=1.0]

Read on dev.to — LLM tag →

• Apache 2.0
• Mistral 3
• Mistral AI
• Mistral Medium 3
• Mistral Medium 3.1
• Mistral Medium 3.5
• Mistral NeMo 12B
• Mistral Small 3.2
• Mistral Small 3.5
• Mistral Small 4
• NVIDIA
• open-mistral-nemo-2407
• Tekken tokenizer

• model release
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →