Nemotron 3.5 Super VL: NVIDIA의 120B 오픈 비전 모델 출시, 5배 주장은 아직 점수판이 없다

N​VIDIA N​emotron 3.5 Super VL shipped on October 15, 2026 — an open 120-billion-parameter vision-language model with open weights, open datasets and training recipes, and a claim attached to it that nobody can check yet. The company's launch post calls it "the fastest open agentic model in its class — 5x faster than the next best," and attributes the number to the Artificial Analysis Intelligence Index leaderboard. As of this writing, Artificial Analysis does not list Super VL at all; its N​emotron 3.5 entries stop at N​emotron 3.5 Lightning. That does not make the claim false. It makes it vendor-reported, and it should be read that way until a third party runs the checkpoint.

The other thing worth knowing up front is a three-day gap. The weights went up on Hugging Face at 2026-10-12T15:32Z — the BF16 post-trained build, the base model and the NVFP4 build all within a minute of each other — and the announcement post did not appear until October 15 at 15:00Z, with the model card's own release-date field still reading 10/15/2026. N​VIDIA uploaded first and announced later. That is now the normal shape of an open-weights release: the repository is the primary source, and the blog post is the commentary.

실제로 출시된 것과 그렇지 않은 것

세 개의 Super VL 체크포인트, 모두 지금 사용 가능합니다:

• N​VIDIA-N​emotron-3.5-Super-VL-120B-A12B-BF16 — the full-precision post-trained model, the reference build and the one to fine-tune from.

• N​VIDIA-N​emotron-3.5-Super-VL-120B-A12B-NVFP4 — the quantized deployment build, produced with N​VIDIA Model Optimizer and quantization-aware distillation from the BF16 model.

• N​VIDIA-N​emotron-3.5-Super-VL-120B-A12B-Base-BF16 — the pre-post-training base, for anyone who wants to run their own post-training rather than adapt N​VIDIA's.

There is a fourth repository, N​VIDIA-N​emotron-3.5-Super-VL-120B-A12B-NVFP4-LiLiCorr, which is not a model you prompt. It is a speculative-decoding draft checkpoint: load it alongside the NVFP4 target and it proposes tokens ahead of the main model. Its own card dates it 10/15/2026 and describes it as intended for LiLiCorr deployments tuned for DGX Spark and low-concurrency data-center work.

Two things did not ship, and each matters. There is no price. N​VIDIA's announcement states the model is available as an N​VIDIA NIM microservice — a production microservice the company says is refreshed every two weeks with security patches and performance work — but publishes no per-token figure anywhere, for NIM or for the third-party platforms carrying it. And there is no independently measured speed number, for the reason above. Everything in the next two sections is N​VIDIA's own reporting.

아키텍처를 한 문단으로, 배포 스토리를 설명해 주기 때문입니다.

Super VL은 Mamba-2–Transformer 하이브리드 전문가 혼합(MoE) 모델입니다. Mamba-2, MoE, 어텐션 레이어가 교차 배치되고, MoE 레이어당 512개의 라우팅된 전문가가 있으며, 여기에 C-RADIO 비전 인코더와 언어 백본에 덧붙여진 비전-언어 커넥터가 더해집니다. 총 120B 파라미터, 토큰당 12B 활성. 컨텍스트는 최대 1M 토큰까지 지원하며, Multi-Token Prediction이 기본으로 내장되어 있는데, 바로 이것을 LiLiCorr 드래프트 체크포인트가 활용합니다. 입력은 텍스트, 이미지, 비디오입니다 — 비디오는 최대 2분 길이의 MP4로, 프레임으로 샘플링됩니다 — 또한 문서는 페이지 이미지와 함께 제공되는 텍스트로 입력할 수 있습니다. 출력은 텍스트뿐입니다.

That shape is why the deployment table is the most useful part of the documentation. N​VIDIA's BF16 build wants 4× H100 or 2× GB200. The NVFP4 build wants 1× DGX Spark, 1× H200, 2× H100, or 1× GB200. The gap between those two lines is the whole reason the quantized checkpoint exists, and it is the difference between "a 120B model you rent" and "a 120B model on the desk."

The benchmarks N​VIDIA published, and the two problems with them

Label all of this vendor-reported. N​VIDIA evaluated Super VL through NeMo Gym and published the recipes, containers, prompts and scoring settings — which is more than most labs do — but N​VIDIA trained the model, chose the baselines and ran the harnesses. Nothing here has been reproduced outside the company.

The agentic numbers are the ones the launch leans on. SWE-bench Verified 71.10, SWE-bench Multilingual 67.72, SWE-bench Pro 51.78, tau2-bench retail 69.24, airline 76.50, telecom 73.19, BrowseComp 78.91, and GraphWalks 89.71 at the over-1M-token setting. On reasoning, GPQA-Diamond 89.52 and HLE without tools 31.79, rising to 46.01 with tools — that last figure, N​VIDIA notes, comes from a separate internal harness rather than the standard path.

On vision, which is the new capability relative to the text-only N​emotron 3.5 Lightning from August: CharXiv-RQ 80.82, MathVision 78.16, V* 85.60, OCR-Reasoning 65.08, OCRBench-v2 at 68.12 English and 59.15 Chinese, MMLongBench-Doc 62.82, OSWorld 52.14, and Video-MME-v2 32.85.

이제 두 가지 문제입니다.

First, the Terminal-Bench 2.1 figure is stated twice and does not agree with itself. The model card reports 66.70. The technical report's summary table reports 60.67. Everything else in those two documents lines up, so this is not two different models — it is one number carrying two values in N​VIDIA's own paperwork, and until the recipes are re-run it is not safe to quote either. That is the sort of thing a single independent replication settles in an afternoon, which is exactly why the absence of one matters this week.

Second, the headline efficiency claim and the headline intelligence claim pull in opposite directions, and only one of them is checkable. N​VIDIA says Super VL reaches accuracy on multimodal agentic tasks "comparable with N​VIDIA's largest reasoning model at roughly one-fifth the parameter count" — N​emotron 3 Ultra 550B A55B has 55B active parameters against Super VL's 12B, a ratio of 4.6×, and the report says Super VL's 51.78 on SWE-bench Pro does surpass Ultra. Fine.

But look at the columns where Super VL is genuinely behind. MMLU-ProX 76.91, against Qwen3.5-122B-A10B at 83.2 and Qwen3.8-Flash-Next at 83.0. AA-Omniscience accuracy 26.20, which is close to the floor of that benchmark and sits below DeepSeek-V4-Flash-0731's 45.0. Video-MME-v2 32.85 against Qwen3.8-Flash-Next at 45.1. BabyVision 29.60 against 46.8. Agentic-terminal at 60.67 or 66.70 against 85.80. Super VL is not the accuracy leader in this weight class on most rows; it is a 12B-active model that N​VIDIA is selling on speed and on the breadth of what it ingests in one call, not on beating everything at everything.

효율 수치들, 그게 사실 흥미로운 것들이죠

지능 비교를 걷어내면, 엔지니어링 수치는 구체적이다:

• Speculative decoding — N​VIDIA's post-training MTP boosting lifts average acceptance length from 4.287 to 4.718 tokens under greedy decoding and 3.787 to 4.612 under probabilistic draft sampling.

• Quantization — the released NVFP4 checkpoint recovers 99.62% of BF16 accuracy as a median across 36 benchmarks, so the deployment build is close to lossless by N​VIDIA's own measure.

• 처리량 — 20턴 에이전트 프록시 워크로드에서 단일 GB200 기준 GPU당 최대 4,856 출력 토큰/s를 달성하면서 사용자당 최소 50 토큰/s를 유지; 동일 조건에서 NVFP4는 BF16 처리량의 2.43배.

• 상대 처리량 — 해당 워크로드 기준 Qwen3.5-122B-A10B 2.34×, Qwen3.8-Flash-Next 1.90×.

• Serving stack — on N​VIDIA Dynamo, N​VIDIA claims up to 8× more concurrent users at 85% lower latency.

그것들은 모두 벤더가 제시한 수치다. 그중 누구든 가장 쉽게 조작할 수 있는 것은 4,856 tokens/s/GPU라는 수치다. 왜냐하면 이는 채점 하네스 뒤에서 실행되는 벤치마크의 속성이 아니라, 특정 GPU와 특정 워크로드의 속성이기 때문이다.

그것을 서빙하는 것, 그리고 두 빌드가 하드웨어에 드는 비용

Both vLLM and SGLang are named as supported runtimes, and N​VIDIA ships working launch configurations: four-way tensor parallelism with three MTP draft tokens on 4× H100 for BF16, two-way with five draft tokens on 2× GB200. Hardware coverage is Hopper and Blackwell, Linux only, and N​VIDIA lists DGX Spark, DGX Station, RTX PRO, RTX and Jetson on the client side — the same silicon spread the model is positioned against.

런칭 당일 이 글을 읽는 팀이 내려야 할 실질적인 결정은 어떤 체크포인트를 받을지입니다. 파인튜닝을 한다면 BF16 빌드가 유일하게 합리적인 출발점입니다. NVFP4 빌드는 이미 양자화 인식 증류를 거쳤기 때문에 추가 학습이 그것과 깔끔하게 결합되지 않기 때문입니다. 서빙을 한다면 NVFP4 빌드가 단일 H200이나 DGX Spark에 들어가며, 중앙값 99.62%의 복구율 덕분에 특정 회귀를 측정해 확인한 경우가 아니라면 이것이 기본값입니다. 베이스 모델을 받는 것은 자체적인 포스트트레이닝 파이프라인을 작성할 계획인 경우에만 의미가 있는데, 그건 별개의 프로젝트입니다.

Super VL의 실제 용도

Read N​VIDIA's own framing and the target is narrow and specific: always-on enterprise agents that digest visual documents. The announcement's example is a web page — extract the text, read the charts, return findings — and the same claim is extended to PDFs, tables and structured documents. The argument for one model instead of a text model plus a separate vision pipeline is context continuity: the document, the data and the decision stay in the same window rather than being handed between systems.

그 주장은 스펙 시트상으로는 설득력이 있다. 컨텍스트 윈도가 1M 토큰이고, 비전 인코더가 사후에 덧붙은 것이 아니라 공동으로 훈련되기 때문이다. 1M 토큰에서 실제 문서와 맞닥뜨렸을 때도 그것이 견디는지는 바로 GraphWalks와 MMLongBench-Doc 수치가 겨냥하는 질문이며, 둘 다 벤더가 보고한 수치다.

N​VIDIA also lists a long adoption roster in the launch post — Scale AI building agents for Skydance and EY, AfterQuery for Hebbia's credit-agreement analysis, AI21 and Alice on cybersecurity, Mercor on coding agents, plus Atlassian, CrowdStrike, Dassault Systèmes, Perplexity, Salesforce and Synopsys among integrators. Those are partnerships named in a press release, which is a statement of intent rather than evidence of production traffic. Treat the roster as a signal about which verticals N​VIDIA is targeting, not as a reliability claim.

프로덕션 경로를 걸지 않고 시험해 보는 방법

이번 주 솔직한 입장은 Super VL이 흥미롭고 문서화도 잘 되어 있지만 전혀 검증되지 않은 오픈 모델이라는 것입니다. 그리고 검증되지 않은 오픈 모델을 평가하는 올바른 방법은 첫날부터 프로덕션 트래픽을 이 모델로 라우팅하는 것이 아닙니다.

Locally, the NVFP4 build on a single H200 or a DGX Spark is the cheap test: pull the checkpoint, run your own document set through it, and compare against whatever vision model you use today. That costs you GPU time and nothing else, and it is the only way to get a number that is not N​VIDIA's. N​VIDIA's own documentation gives you one free head start: the model card and the technical report both specify temperature 1.0 and top_p 0.95 for text, and temperature 0.6 with top_p 0.95 for vision — a starting point you would otherwise spend a day rediscovering.

If you would rather test it behind an API, the availability line matters: Super VL is on Hugging Face, as N​VIDIA's own NIM microservice, and through several third-party platforms. N​VIDIA has published no price for any of those paths, so any per-token figure you see this week is set by the platform, not by the vendor. On the OrcaRouter side, N​emotron 3.5 Super VL is not on our catalogue — we do not serve it today, so there is nothing here to try it through yet, and we would rather say that than imply otherwise. Everything we do serve sits behind one key across 200-plus models at provider list price with 0% markup, and where a model is available from more than one provider, automatic failover across providers gives you a way to route to a new model on a fraction of traffic and let it fall back to the incumbent when it stumbles. That is the mechanism this kind of launch wants; it is not an offer to serve this particular checkpoint.

One more piece of the release worth a look regardless of whether you want the model: N​VIDIA published its post-training datasets alongside it — spans covering agentic tool use, finance, instruction following, long context, fairness and vision-language understanding — plus the RL environments in NeMo Gym. The dataset release is often more reusable than the checkpoint, and it is the part of this launch that costs N​VIDIA the most to give away.

볼 만한 것

이 글을 바꿀 네 가지가 있는데, 대략 그것들이 얼마나 중요한지에 따른 순서로 말이다.

Artificial Analysis든 다른 누구든, 5배라는 주장이 함축하는 Super VL 행을 산출하거나 아니면 산출하기를 거부하는 최초의 독립적 벤치마크. 그것이 존재하기 전까지, 이번 출시의 핵심 효율성 주장은 리더보드 이름만 붙은 보도자료 속 한 문장에 불과하며, 그 단서는 그 수치와 한 호흡에 함께 있어야 한다.

A price. N​VIDIA has told you the hardware, the runtimes and the licence, and withheld the one figure a buyer needs to compare this against a closed model. When NIM pricing appears, the comparison becomes arithmetic instead of positioning.

The Terminal-Bench 2.1 discrepancy — 66.70 on the model card against 60.67 in the technical report. Whichever is right, one of N​VIDIA's two primary documents is wrong, and a corrected re-run would settle several other rows at the same time.

And the walk from "available now" to "actually served." Super VL's benchmark rows compare it against Qwen3.5-122B-A10B, Qwen3.8-Flash-Next, DeepSeek-V4-Flash-0731, GLM-5.3-Flash and N​emotron 3 Ultra 550B A55B — all of which are mature enough to have measured throughput, real prices and production track records. A 12B-active model that ingests video, 1M-token documents and tool calls in one pass is a genuinely useful shape. Whether it earns a routing rule is a question about latency and price stability under load, and N​VIDIA has published the first of those only on its own hardware, under its own workload, in its own units.

벤치마크는 낡아가고 가격은 변동하지만, 실패 기록은 그렇지 않습니다: 이 계열의 모델과 관련된 문서화된 AI 사고와 에이전트 실패는 OrcaRouter AI Incident Archive에 수집되어 있으며, 각 항목은 자체 출처, 심각도, 신뢰도 등급, 이의 제기 플래그를 갖추고 있습니다.