{"slug": "nemotron-3-super", "name": "Nemotron 3 Super", "category": "ai-models", "category_name": "AI models", "description": "", "recommenders": 3, "commenters": 6, "warners": 0, "score": 3.261, "rising": -0.98, "first": "2026-03-22", "latest": "2026-07-31", "why": "we're using Nemotron-3-super with incredible success", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 3, "neutral": 0, "compare": 3, "warn": 0}, "spark": [0, 0, 0, 0, 0, 1, 0, 1, 1, 0, 0, 0], "listed": true, "loved": [], "caveat": "", "short_quote": "we're using Nemotron-3-super with incredible success", "kind": "", "icon": true, "family": "nemotron", "domain": "nvidia.com", "github_repo": null, "official_url": "https://developer.nvidia.com/nemotron", "pricing": "", "aliases": ["Nemotron 3 Super", "Nemotron-3-Super"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 785968081, "item_id": 49123005, "url": "https://news.ycombinator.com/item?id=49123005", "by": "markab21", "date": "2026-07-31", "version": "3", "stance": "recommend", "quote": "we're using Nemotron-3-super with incredible success", "excerpt": "We use it at a moderate scale, self-hosted on B300 hardware. It's great :D QA analysis of voice transcriptions. Napkin math: we operate at 2-5% of the cost of running on Equiv Frontier, though this changes near-weekly because pricing is so volatile. It took us about a month to get the inference configured to achieve these numbers. But if you can get your hands on a pair of B300 GPUs and the context works, it's untouc", "context": "self-hosts at moderate scale on B300 GPUs", "use_case": "lighter inference tasks", "switched_from": "", "story": "DeepSeek-V4-Flash Update"}, {"id": 777972386, "item_id": 48623274, "url": "https://news.ycombinator.com/item?id=48623274", "by": "coder543", "date": "2026-06-21", "version": "3", "stance": "recommend", "quote": "There are equally open, much more useful models out there:", "excerpt": "Is a recipe useful if no one likes it? There are equally open, much more useful models out there: https://artificialanalysis.ai/?models=nvidia-nemotron-3-ultra-550b-a55b%2Cnvidia-nemotron-3-super-120b-a12b%2Colmo-3-1-32b-instruct%2Colmo-3-1-32b-think%2Ck2-think-v2%2Capertus-8b-instruct%2Capertus-70b-instruct#intelligence", "context": "", "use_case": "", "switched_from": "", "story": "Apertus \u2013 Open Foundation Model for Sovereign AI"}, {"id": 761932673, "item_id": 47620792, "url": "https://news.ycombinator.com/item?id=47620792", "by": "mratsim", "date": "2026-04-02", "version": "3", "stance": "recommend", "quote": "Nemotron-3-Super: 14070 tok/s prefill, ~194.5 tok/s generation", "excerpt": "In my case with vLLM on dual RTX Pro 6000 gpt-oss-120b: (unknown prefill), ~175 tok/s generation. I don't remember the prefill speed but it certainly was below 10k Nemotron-3-Super: 14070 tok/s prefill, ~194.5 tok/s generation. (Tested fresh after reload, no caching, I have a screenshot.) Nemotron-3-Super using NVFP4 and speculative decoding via MTP 5 tokens at a time as mentioned in Nvidia cookbook: https://docs.nvi", "context": "", "use_case": "", "switched_from": "", "story": "Google releases Gemma 4 open models"}]}], "caveats": [], "who": [{"by": "markab21", "context": "self-hosts at moderate scale on B300 GPUs", "url": "https://news.ycombinator.com/item?id=49123005", "date": "2026-07-31"}], "tags": [], "use_cases": ["lighter inference tasks", "local agentic inference"], "compared": [{"slug": "gpt-oss", "name": "gpt-oss", "co": 3, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=47616361", "title": "Google releases Gemma 4 open models", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=49119559", "title": "DeepSeek-V4-Flash Update", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48622778", "title": "Apertus \u2013 Open Foundation Model for Sovereign AI", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=47419391", "title": "Mamba-3", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {}, "2026-03": {"compare": 1}, "2026-04": {"compare": 2, "recommend": 1}, "2026-05": {}, "2026-06": {"recommend": 1}, "2026-07": {"recommend": 1}, "2026-08": {}, "2026-09": {}, "2026-10": {}}, "versions": [], "version_unnamed": 0, "family_ref": {"slug": "nemotron", "name": "Nemotron"}, "runs_on": null, "lines": [], "quotes": [{"id": 785968081, "item_id": 49123005, "url": "https://news.ycombinator.com/item?id=49123005", "by": "markab21", "date": "2026-07-31", "version": "3", "stance": "recommend", "quote": "we're using Nemotron-3-super with incredible success", "excerpt": "We use it at a moderate scale, self-hosted on B300 hardware. It's great :D QA analysis of voice transcriptions. Napkin math: we operate at 2-5% of the cost of running on Equiv Frontier, though this changes near-weekly because pricing is so volatile. It took us about a month to get the inference configured to achieve these numbers. But if you can get your hands on a pair of B300 GPUs and the context works, it's untouc", "context": "self-hosts at moderate scale on B300 GPUs", "use_case": "lighter inference tasks", "switched_from": "", "story": "DeepSeek-V4-Flash Update"}, {"id": 777972386, "item_id": 48623274, "url": "https://news.ycombinator.com/item?id=48623274", "by": "coder543", "date": "2026-06-21", "version": "3", "stance": "recommend", "quote": "There are equally open, much more useful models out there:", "excerpt": "Is a recipe useful if no one likes it? There are equally open, much more useful models out there: https://artificialanalysis.ai/?models=nvidia-nemotron-3-ultra-550b-a55b%2Cnvidia-nemotron-3-super-120b-a12b%2Colmo-3-1-32b-instruct%2Colmo-3-1-32b-think%2Ck2-think-v2%2Capertus-8b-instruct%2Capertus-70b-instruct#intelligence", "context": "", "use_case": "", "switched_from": "", "story": "Apertus \u2013 Open Foundation Model for Sovereign AI"}, {"id": 761932673, "item_id": 47620792, "url": "https://news.ycombinator.com/item?id=47620792", "by": "mratsim", "date": "2026-04-02", "version": "3", "stance": "recommend", "quote": "Nemotron-3-Super: 14070 tok/s prefill, ~194.5 tok/s generation", "excerpt": "In my case with vLLM on dual RTX Pro 6000 gpt-oss-120b: (unknown prefill), ~175 tok/s generation. I don't remember the prefill speed but it certainly was below 10k Nemotron-3-Super: 14070 tok/s prefill, ~194.5 tok/s generation. (Tested fresh after reload, no caching, I have a screenshot.) Nemotron-3-Super using NVFP4 and speculative decoding via MTP 5 tokens at a time as mentioned in Nvidia cookbook: https://docs.nvi", "context": "", "use_case": "", "switched_from": "", "story": "Google releases Gemma 4 open models"}, {"id": 761924929, "item_id": 47620308, "url": "https://news.ycombinator.com/item?id=47620308", "by": "evilduck", "date": "2026-04-02", "version": "3", "stance": "compare", "quote": "nemotron-3-super is at best 260 tokens/s PP.", "excerpt": "In terms of ability, maybe, in terms of speed, it's not even close. Check out the Prompt Processing speeds between them: https://kyuz0.github.io/amd-strix-halo-toolboxes/ gpt-oss-120b is over 600 tokens/s PP for all but one backend. nemotron-3-super is at best 260 tokens/s PP. Comparing token generation, it's again like 50 tokens/sec vs 15 tokens/sec That really bogs down agentic tooling. Something needs to be catego", "context": "", "use_case": "local agentic inference", "switched_from": "", "story": "Google releases Gemma 4 open models"}, {"id": 761897376, "item_id": 47618586, "url": "https://news.ycombinator.com/item?id=47618586", "by": "kcb", "date": "2026-04-02", "version": "3", "stance": "compare", "quote": "Nemotron 3 Super was released recently. That's a direct competitor to gpt-oss-120b.", "excerpt": "Nemotron 3 Super was released recently. That's a direct competitor to gpt-oss-120b. https://developer.nvidia.com/blog/introducing-nemotron-3-super-an-open-hybrid-mamba-transformer-moe-for-agentic-reasoning/", "context": "", "use_case": "", "switched_from": "", "story": "Google releases Gemma 4 open models"}, {"id": 759757696, "item_id": 47484856, "url": "https://news.ycombinator.com/item?id=47484856", "by": "lambda", "date": "2026-03-22", "version": "3", "stance": "compare", "quote": "Nemotron 3 Super doesn't perform quite as well on benchmarks as the similarly sized Qwen3.5 122B A10B model, but it goes faster and is cheaper to run.", "excerpt": "It mostly trades some potential performance loss for speed, especially at longer contexts. Nemotron 3 Super doesn't perform quite as well on benchmarks as the similarly sized Qwen3.5 122B A10B model, but it goes faster and is cheaper to run. https://artificialanalysis.ai/?models=gpt-oss-120b%2Cmistral-small-4%2Cnvidia-nemotron-3-super-120b-a12b%2Cqwen3-5-122b-a10b Now, you're not exactly comparing apples to apples th", "context": "", "use_case": "", "switched_from": "", "story": "Mamba-3"}]}