NVIDIA's Nemotron 3.5 Super VL, a 120B open-weight multimodal model, hits Together AI
togethercompute · x · 2026-10-15
Together AI now serves NVIDIA Nemotron 3.5 Super VL, an open-weight multimodal reasoning model for visual AI agents. Key details:
- 120B MoE (12B active) with hybrid Mamba-Transformer architecture
- Unified text, image, document and video understanding with reasoning and tool use, preserving visual context across steps
- Multi-token prediction for faster generation; NVFP4 KV cache for more concurrent long-context sessions
- Open weights, customizable via post-training on proprietary data
Positioned as the high-accuracy multimodal tier alongside Nemotron 3.5 Lightning (high-volume text) and Nemotron 3 Ultra (hardest text-only reasoning).
More from Infra
- Muse-Spark open-weights drop: 816B-A42B MoE model gets 359GB MLX 3-bit build — ns123abc · 2026-10-16
- vLLM ships Day-0 LoRA RL for Muse Spark Open, lifting AIME from 0.56 to 0.85 — vllm_project · 2026-10-16
- DSpark draft model for Muse Spark Open proposes 7 tokens per pass with 23-42% acceptance — vllm_project · 2026-10-16
- Cloudflare Stream's WebRTC ultra-low-latency live streaming is now Generally Available — ritakozlov · 2026-10-16
- YC-backed Overshoot launches inference optimization, claims $5M saved for AI teams — ycombinator · 2026-10-16
- Meta's 816B open-weight Muse Spark Open hits Groq preview at 350+ tokens/sec — GroqLLC · 2026-10-16