{"slug": "nemotron-3-ultra", "title": "Nemotron 3 Ultra", "summary": "NVIDIA released Nemotron 3 Ultra, a 550-billion-parameter open-weight MoE model with hybrid Mamba-2 and Transformer architecture, under the Linux Foundation's OpenMDW-1.1 license. The model requires server-class hardware (minimum 8x H200 GPUs) and achieves top scores among US open-weight models, though it trails Kimi K2.6 on hard reasoning and coding benchmarks.", "body_md": "# Nemotron 3 Ultra\n\nMoE premier550B total, 55B active per token (MoE) - NVIDIA's from-scratch hybrid Mamba-2 + Transformer with LatentMoE routing and 5-token Multi-Token Prediction (MTP) speculative decode. 262K default context (up to 1M with env flags); RULER@1M 94.7. Genuinely open-weight under the Linux Foundation's OpenMDW-1.1 license, downloadable on HuggingFace (BF16 + NVFP4 checkpoints). Server-class only: NVIDIA's stated minimum is 8x H200, 16x H100, or 8x B200/GB200 - BF16 weights are ~1.1TB and the NVFP4 checkpoint ~275GB, so it does not fit a single RTX 5090, RTX 6000 Ada, or Mac Studio M3/M4 Ultra. Not workstation-fittable, so cloud-only for nearly everyone despite open weights. Benchmarks are NVIDIA self-reported (MMLU-Pro 86.8, GPQA 87.0, SWE-bench Verified 70.7); independent eval (Artificial Analysis, Intelligence Index 47.7, #9 of 89) calls it the leading US open-weight model but ~6 points behind Kimi K2.6 on hard reasoning and coding. Treat the vendor throughput figures as claims pending community replication.\n\n- 550.0B\n- 262k\n- other\n- Jun 2026\n\n## Scores\n\n## Or run it in the cloud\n\nLive per-provider pricing, throughput and uptime - refreshed 3 days ago via OpenRouter. Click a column to sort.\n\n| Provider | Type | Input $/M | Output $/M | Cache $/M | Tok/s | Latency | Uptime | Value |\n|---|---|---|---|---|---|---|---|---|\n| API | 0.50 | 2.20 | - | - | - | - | cheapest |\n\nDefault order: throughput among 95%+ uptime providers, then latency; subscriptions last. Sort by any column. Subscription rows show $/mo in the Value column - per-token columns are \"-\". Affiliate links are marked sponsored / nofollow. Confirm current pricing on the provider's site before committing.\n\n[Detailed API pricing page + JSON endpoint →](/models/nemotron-3-ultra/pricing)\n\n## Inference cost over time\n\nData accumulates from the first daily sync - longer ranges populate over time. Prices come from OpenRouter snapshots, not a historical API.", "url": "https://wpnews.pro/news/nemotron-3-ultra", "canonical_source": "https://tokenstead.ai/models/nemotron-3-ultra", "published_at": "2026-07-09 14:12:02+00:00", "updated_at": "2026-07-09 15:09:45.890895+00:00", "lang": "en", "topics": ["large-language-models", "ai-products", "ai-infrastructure", "ai-research"], "entities": ["NVIDIA", "Nemotron 3 Ultra", "Linux Foundation", "OpenMDW-1.1", "HuggingFace", "Artificial Analysis", "Kimi K2.6", "OpenRouter"], "alternates": {"html": "https://wpnews.pro/news/nemotron-3-ultra", "markdown": "https://wpnews.pro/news/nemotron-3-ultra.md", "text": "https://wpnews.pro/news/nemotron-3-ultra.txt", "jsonld": "https://wpnews.pro/news/nemotron-3-ultra.jsonld"}}