Recent events
Recent events tied to this brand in public sources.
Back to profile5 entries
2026
Orange Cyberdefense and Mistral AI agree to use Mistral Large 4
Orange Cyberdefense reached an agreement with Mistral AI to use the model’s cybersecurity version. The initial rollout is intended for selected services for enterprise and public-sector customers, including testing customer defenses for vulnerabilities and speeding up intrusion investigations.
2026
Mistral AI launches voice agent SDK
Mistral AI launched a voice agent SDK based on LiveKit, integrating transcription, inference, tool calling and speech generation into a real-time streaming pipeline.
2026
Mistral AI launches flagship model Mistral Large 4 (Le Chonk) with API preview and planned open-weight release
French AI startup Mistral AI released its new flagship model Mistral Large 4, nicknamed "Le Chonk", with 1.05 trillion total parameters using a mixture-of-experts architecture. The model became available for public preview via its API on October 6, 2026, and a full open-weight version is expected on October 31, 2026. The model's API is priced at USD 1.36 per million input tokens and USD 4.18 per million output tokens.
2026
Mistral AI releases Mistral Large 4, a 1-trillion-parameter multimodal model
On Tuesday, French AI lab Mistral AI released Mistral Large 4 (ML4), a large multimodal model with 1 trillion parameters nicknamed Le Chonk, aiming to compete with American and Chinese rivals. The model is currently accessible only via a public guardrail endpoint and is not yet open-weight; Mistral plans to make its weights available in three weeks after safety testing is complete.
2026
Mistral AI launches Agentic Search
Mistral AI launched Agentic Search, a multi-step retrieval technology built on existing enterprise search indexes that gives models five file-system-like tools (search, open, navigate, read, grep) to iteratively inspect and compare long complex documents. In FinanceBench tests on 368 SEC filings averaging about 147 pages, accuracy rose about threefold with the search loop alone and up to 86% with the full navigation toolchain, while p90 latency and token consumption fell. It supports cloud and on-premises deployment and can be embedded in Studio and Vibe Libraries or integrated via the Mistral Search Toolkit.