Together AI – Weekly Recap

Together AI is the focus of this weekly summary of notable developments, as the company broadened its AI infrastructure capabilities and deepened its role in advanced research and government initiatives. Across product launches, partnerships and thought-leadership efforts, Together AI continued to position itself as a key player in open, high-performance AI for agentic and multimodal workloads.

The company promoted an upcoming webinar on deploying the DeepSeek-V4 model for long-context, agentic workflows, emphasizing the technical challenge of serving such architectures at speed and scale. By highlighting its VP of Kernels and senior inference and research staff, Together AI underscored its expertise in low-level optimization that could appeal to enterprise customers demanding reliable, high-throughput agentic applications.

Together AI also expanded its fine-tuning offering via a partnership with Adaptive Data, integrating Together Fine-Tuning directly into Adaptive Data’s platform. This workflow links curated and optimized datasets to fine-tuned open models hosted on Together AI, aiming to tighten the loop from data preparation through evaluation and production deployment for AI-native teams.

On the model catalog side, the company added Qwen3.6-Plus to its serverless AI Native Cloud, targeting agentic coding, multimodal reasoning and long-horizon workflows with a 1 million token context window. Transparent, usage-based pricing for inputs and outputs signals a push to monetize advanced inference workloads and strengthen Together AI’s appeal to developers building production-grade agent systems.

Together AI further expanded its multimodal capabilities by integrating NVIDIA’s Nemotron 3 Nano Omni model into its production inference platform, supporting audio, video, image, document and text within a single architecture. The firm emphasized a hybrid Mamba-Transformer mixture-of-experts design enabling higher throughput, fully managed scaling and a security-focused, zero-trust approach aimed at risk-sensitive enterprises.

Beyond commercial offerings, Together AI joined the U.S. Department of Energy’s Genesis Mission, a decade-long initiative to apply AI across 17 national labs with an emphasis on open, researcher-controllable models. The company also highlighted participation in an ICLR workshop on neural reasoning, signaling continued investment in frontier research that could reinforce its long-term technological differentiation in AI infrastructure.

Collectively, these initiatives suggest a week of meaningful progress for Together AI across infrastructure, partnerships, public-sector engagement and research, potentially enhancing its competitive positioning and supporting future usage-driven growth without materially changing near-term risk factors.