Oct 7, 2026
ManyPress

Advertisement

Technology

French AI startup Mistral has released Mistral Large 4, a 1 trillion-parameter model designed to rival top-tier open-weight and proprietary AI systems.

ManyPress

ManyPress

ManyPress Editorial

3 min readSource:The Hindu Technology, Ars Technica
Mistral Unveils 'Le Chonk' AI Model to Compete With Global Rivals

Key facts

  • •Mistral Large 4 is a 1 trillion-parameter model with 49 billion active parameters.
  • •The model was trained from scratch using 3,800 NVIDIA Grace Blackwell GPUs in Europe.
  • •Mistral reported a Coding Agent Index score of 49.8% for the new model.
  • •The company raised $3.3 billion in a funding round in September 2026, reaching a $24 billion valuation.
  • •Public weights for the model are scheduled for release by the end of October 2026.

French AI startup Mistral announced the release of its Mistral Large 4 model, nicknamed 'Le Chonk,' on Tuesday, October 6, 2026. The company describes the 1 trillion-parameter multimodal model as its most capable to date, positioning it as a direct competitor to both closed-source U.S. models and open-weight models from Chinese firms like Moonshot and DeepSeek.

By the numbers

1 trillion
total parameters in Mistral Large 4
49 billion
active parameters in the model
49.8%
Coding Agent Index score
$3.3 billion
amount raised in September funding round
$24 billion
company valuation

Technical Capabilities and Training

Le Chonk features 49 billion active parameters and is designed for proficiency in coding, agentic workflows, and multimodal understanding. Mistral reports that the model is fluent in over 160 languages and excels in specialized fields including cybersecurity, finance, law, and visual grounding. According to the company, the model was trained from scratch using 3,800 NVIDIA Grace Blackwell GPUs located in European data centers, a process Mistral highlights to distinguish its development from Chinese models accused of relying on distillation techniques.

Market Positioning and Availability

Mistral CEO Artur Mensch stated that the model outperforms certain Chinese rivals in areas such as cybersecurity, challenging the narrative that Europe cannot compete in the global AI race. The company claims its Coding Agent Index score of 49.8% surpasses that of the DeepSeek V4 Pro 0813 and Qwen3.8 Max. While currently available in a public preview, the company plans to release the open weights by the end of October. During this interim period, Mistral is conducting red-teaming exercises with cybersecurity leaders and state authorities to test the model's capabilities.

Business Strategy and Growth

Mistral differentiates its business model by offering open-weight alternatives to the proprietary, closed-access systems developed by U.S. labs like OpenAI and Anthropic. The company generates revenue through pay-as-you-go cloud fees and custom engineering services. This release follows a period of significant growth for the French firm, which recently secured $3.3 billion in funding, resulting in a $24 billion valuation. Mistral leadership indicated that this launch marks the beginning of a faster development cycle for future specialized models.

Timeline

  1. July 2026
    Chinese AI company Moonshot launched its 2.8 trillion-parameter Kimi K3 model.
  2. September 2026
    Mistral secured a $3.3 billion funding round.
  3. October 6, 2026
    Mistral announced the public preview of Mistral Large 4.

Advertisement

This article was independently rewritten by ManyPress editorial AI from reporting originally published by The Hindu Technology, Ars Technica.

Technology