Brief

Meta AI has launched Muse Spark, a new natively multimodal reasoning model designed for personal superintelligence applications. This model integrates visual understanding, tool use, and multi-agent orchestration, with a special 'Contemplating mode' for advanced reasoning. Meta AI has also invested in infrastructure, including the Hyperion data center, and claims significant improvements in training efficiency compared to their previous model, Llama 4 Maverick. AI

Together AI has enhanced its fine-tuning platform to support a wider array of large language models, including recent releases from DeepSeek, Qwen, and Meta, alongside OpenAI's gpt-oss. The platform now offers expanded context lengths, up to 131k tokens for some models, at no additional cost, facilitating tasks like long-document processing and complex code editing. Separately, Together AI researchers have explored LLM behavior using minimal, topic-neutral prompts to uncover inherent model preferences, finding that GPT-OSS favors programming and math, Llama leans literary, DeepSeek often produces religious content, and Qwen tends toward multiple-choice questions. AI