AI competition is accelerating at an unprecedented pace. Following Meta's release of its new open-source Llama 3.1 model, French AI startup Mistral AI has entered the fray.
Mistral AI has announced the next generation of its flagship open-source model: Mistral Large 2. This model boasts 123 billion parameters and rivals OpenAI and Meta's latest cutting-edge models in code generation, mathematics, and reasoning.
Hot on the heels of Llama 3.1 405B, the release of Mistral Large 2 has energized the open-source large model field with its key feature: "big enough."
Although Mistral Large 2 has fewer parameters than Llama 3.1's 405 billion, their performances are comparable. In several benchmarks, it competes with GPT-4o and Anthropic's Claude 3.5 Sonnet.
