nvidia
NVIDIA: Nemotron 3.5 Super VL
nvidia/nemotron-3.5-super-vl
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
nvidia/nemotron-3.5-super-vl is NVIDIA's open vision-language mixture-of-experts model: a Mamba-2 and Transformer hybrid with 120 billion total and 12 billion active parameters, plus a C-RADIO vision encoder. Per NVIDIA's model card, it accepts text, images, document pages, GUI screenshots, and short video, returns text, supports up to roughly one million tokens of context, configurable reasoning, tool calling, and native multi-token speculative decoding. OpenRouter lists it as a hosted endpoint; NVIDIA reports broad benchmark coverage, though results are vendor-run and need independent confirmation.[1][2]
Strengths
- NVIDIA's documentation reports a hybrid Mamba-2/Transformer MoE design with 12 billion active parameters and a context window of up to about one million tokens, aimed at long-context efficiency.[2]
- Multimodal scope per the model card: text, images, document page images, GUI screenshots, and video input, with tool calling and an adjustable reasoning mode.[1][2]
- Open weights under the OpenMDW-1.1 license with commercial use permitted; the BF16 release targets customization and domain adaptation, and native MTP speculative decoding is built in.[2]
Limitations
- Nearly all capability and benchmark evidence is vendor-reported; NVIDIA notes some evaluations used internal scaffolding, and OpenRouter's third-party aggregates are sparse, so workload-level validation is essential.[1][2]
- The reference BF16 deployment recipe requires high-end multi-GPU NVIDIA hardware (Hopper or Blackwell), and serving configs cap media per request and target short MP4 clips.[2]
- Knowledge cutoff is not stated, listed language coverage is limited to six languages, and this evidence pack lacks independent evaluations beyond catalog data.[1][2]
Where to use it
- Document and screen understanding agents — for example, parsing page images or GUI screenshots — where tool calling and long context matter (suggested, not a proven result).[1][2]
- Long-context software-engineering or agentic workflows, given the card's reported software-engineering mid-training and coding benchmark results.[2]
- Teams wanting to fine-tune or domain-adapt an open vision-language model using the BF16 reference weights.[2]
Sources
- [1] OpenRouter catalog · NVIDIA: Nemotron 3.5 Super VLSources checked: Nov 13, 2026
- [2] Model documentation · NVIDIA: Nemotron 3.5 Super VLSources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
NVIDIA Nemotron 3.5 Super VL is an open multimodal mixture-of-experts model from NVIDIA, with 12B active parameters out of 120B total. It takes text, images, documents, GUI screenshots, and video...
Specifications
- Model ID
- nvidia/nemotron-3.5-super-vl
- Canonical ID
- nvidia/nemotron-3.5-super-vl-20261015
- Creator
- nvidia
- Added to OpenRouter
- 15.10.2026
- Context length
- 1,048,576
- Modality
- text+image+video->text
- Input modalities
- text, image, video
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $0.150 / out $0.470
- Knowledge cutoff
- —
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | CC — below the B band | reasoning: optional · parameters: include_reasoning, reasoning |
| Coding | No measurement | — |
| Agents | No measurement | parameters: structured_outputs, tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: optional · parameters: include_reasoning, reasoning |
| Document analysis | No measurement | — |
| Vision | No measurement | input: image |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | input: video |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Artificial Analysis | Intelligence index | 13.6 | CC — below the B band | 13.11.2026 |
| Design Arena | models · 3dwin rate 30.2% · rank 84 | 1,116Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · codecategorieswin rate 32.4% · rank 93 | 1,140Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · datavizwin rate 26.1% · rank 106 | 1,085Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · gamedevwin rate 27.1% · rank 106 | 1,078Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · uicomponentwin rate 33% · rank 82 | 1,154Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · websitewin rate 34.8% · rank 88 | 1,160Elo | CC — below the B band | 13.11.2026 |
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| Darkbloomnvfp4 | 262,144 | $0.060 | $0.350 | — | — | 100% |
| Io Netnvfp4 | 262,144 | $0.090 | $0.300 | — | — | 100% |
| Crusoebf16 | 1,048,576 | $0.100 | $0.400 | — | — | 100% |
| CoreWeavenvfp4 | 1,048,576 | $0.150 | $0.470 | — | — | 100% |
| PrimeIntellectnvfp4 | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| Together | 262,144 | $0.150 | $0.500 | — | — | 100% |
| DeepInfrafp4 | 262,144 | $0.150 | $0.500 | — | — | 100% |
| Venicefp4 | 262,144 | $0.200 | $0.500 | — | — | — |
| BaseTenfp4 | 1,048,576 | $0.200 | $0.500 | — | — | 100% |
| Nebiusfp4 | 1,048,576 | $0.300 | $0.900 | — | — | 100% |
Provider data retrieved: 13.11.2026