🔍 Read the full analysis: What Makes Nemotron 3.5 Super VL’s Metrics Stand Out? on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
NVIDIA released Nemotron 3.5 Super VL on October 15, 2026. Artificial Analysis gave it an Intelligence Index score of 26. The model is fast, less costly per benchmark task than its predecessor and commercially usable under its stated license, but several competing models score higher for similar or lower task costs.
The model has 120 billion total parameters, with 12 billion active in its mixture-of-experts architecture, and accepts text and images while producing text. Its listed context window is 1 million tokens. NVIDIA released the weights in NVFP4, a four-bit format. The source describes the license, OpenMDW-1.1, as permitting commercial use.
Artificial Analysis measured a cost of $0.25 per Intelligence Index task. That compares with $1.64 per task for Nemotron 3 Super, according to the source. The new model also scored 26.5 in the specification table, while the article rounds that result to 26 in its headline comparisons. The reported reduction in task cost is attributed in part to fewer output tokens per task and inexpensive cached input.
Speed depends on where the model is served. A pre-release DeepInfra endpoint produced about 630 output tokens per second on 100,000-token prompts, a result Artificial Analysis described as its fastest measured at that prompt length among models scoring at least 20. Across six providers, however, the model page reported median output speeds of 196 tokens per second on standard prompts and 307 on 100,000-token prompts. Median time to the first token was 0.89 seconds, but the source says the model’s reasoning can mean about 11 seconds before the first answer token.
Infographic 01 · What 25 cents buys
Cheap and fast. Not the most intelligence per dollar.
Nemotron 3.5 Super VL costs $0.25 per Intelligence Index task and scores 26. Several open models with MIT licenses score 13 to 20 points more for the same money or less.
Intelligence Index vs. cost to run it
Up and to the left is better. The dashed line marks Super VL’s $0.25. Hover a dot for license and values.
Not plotted: K2 Horizon 375B A23B (UAE, index 30.5, listed at $0.00) and Nemotron 3.5 Bolt 74B-A9B (index 21.4, no API price).
License matters here. Muse Spark Open (Meta, 38.9) scores higher than Super VL but ships under the Muse Model License, which allows commercial use with restrictions. Super VL’s OpenMDW-1.1 license allows commercial use.
Show the data as a table (20 models)
| Model | Creator | License | Index | Cost / task |
|---|---|---|---|---|
| MiMo-V2.6-Pro | Xiaomi | MIT | 46.3 | $0.13 |
| GLM-5.3 | Z AI | GLM-5.3 License (restricted) | 44.8 | $2.01 |
| Step 5 | StepFun | StepFun Model License (restricted) | 43.7 | $1.03 |
| Kimi K3 | Kimi | Kimi K3 License (restricted) | 43.6 | $2.00 |
| Claude Haiku 5.5 (max) | Anthropic | Proprietary | 43.4 | $0.21 |
| GLM-5.3-Flash | Z AI | MIT | 41.8 | $0.28 |
| Ling 3.1 Flash | InclusionAI | MIT | 41.1 | $1.00 |
| Qwen3.8 Flash Next | Alibaba | Qwen Community License (restricted) | 39.8 | $0.37 |
| DeepSeek V4.1 Flash | DeepSeek | MIT | 39.5 | $0.27 |
| Muse Spark Open | Meta | Muse Model License (restricted) | 38.9 | $0.84 |
| GPT-6 Luna (max) | OpenAI | Proprietary | 38.1 | $0.07 |
| MiMo-V2.6-Flash | Xiaomi | MIT | 37.9 | $0.06 |
| DeepSeek V4 Pro | DeepSeek | MIT | 36.0 | $0.67 |
| Qwen3.8 27B | Alibaba | Apache 2.0 | 33.7 | $1.01 |
| MiniMax-M3 | MiniMax | MiniMax Community License (restricted) | 29.2 | $0.51 |
| Nemotron 3.5 Super VL | NVIDIA | OpenMDW-1.1 | 26.5 | $0.25 |
| Inkling Small | Thinking Machines | Apache 2.0 | 25.7 | $0.09 |
| Nemotron 3 Ultra 550B | NVIDIA | OpenMDW | 22.9 | $0.60 |
| Muse Glimmer | Meta | Apache 2.0 | 17.5 | $0.05 |
| Nemotron 3 Super (previous) | NVIDIA | NVIDIA Open Model License | 12.8 | $1.64 |
Source: Artificial Analysis, Nemotron 3.5 Super VL model page and model data for the comparison models, read 16 Oct 2026; launch analysis (15 Oct 2026). Intelligence Index v4.3.2, ten evaluations. Cost per task = total API cost to run the Intelligence Index divided by its tasks. “Restricted” = Artificial Analysis lists commercial use as allowed with conditions.
Powered by Thorsten Meyer AI
Speed Comes With a Value Trade-Off
For operators, the figures point to a model whose case rests on throughput, latency and deployment control rather than leading intelligence per dollar. The source’s comparisons put Xiaomi’s MiMo-V2.6-Pro at a score of 46 for $0.13 per task, and MiMo-V2.6-Flash at 38 for $0.06. Inkling Small is listed at 26 for $0.09. Those figures suggest Super VL’s benchmark-task price is not the lowest among models with similar or higher scores.
The model may still suit teams that prioritize running weights on their own NVIDIA hardware, commercial-use terms, or measured performance on their own workloads. But the provider speed spread is substantial: the reported range runs from 106 tokens per second at the fifth percentile to 534 at the 95th percentile. A result from one endpoint is not a reliable forecast of every deployment’s speed, and the benchmark average does not establish how well a model will perform on a particular company’s tasks.
Cost also depends on usage patterns. The source estimates that about 89% of the $0.25 task cost comes from input, with cache writes accounting for 69% and output costing less than three cents. For systems that repeatedly process long documents or code repositories, input volume and caching could matter more to the bill than the headline output-token rate.
NVIDIA Nemotron 3.5 Super VL AI model
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How the Benchmark Compares Models
The Intelligence Index is an Artificial Analysis benchmark, not a direct measure of every real-world use. The source says Nemotron 3.5 Super VL ranks first among 67 open-weight models in its size class, defined as models with 40 billion to 150 billion total parameters. Counting open-weight models across sizes, it places around 18th. The size-class ranking should not be read as first among all open models.
ThorstenMeyerAI.com reports improvements over Nemotron 3 Super in several benchmark areas. GDPval-AA, which assesses economically useful work, rose to 1,162 Elo from 490. AA-Briefcase was reported at 999 Elo, compared with 876 for Nemotron 3 Ultra. AutomationBench-AA reached 14.8%, versus 3.0% for Ultra; the source notes that the new score remains low in absolute terms.
The source also reports an AA-Omniscience hallucination rate of 36%, down from 87% for the predecessor. That is a benchmark result, not a guarantee that the model will avoid unsupported answers in production. License comparisons also need care: the article says Super VL leads among permissively licensed US open-weight models, while Meta’s Muse Spark Open has a more restrictive license despite a higher reported score.
“Super VL is a very good engine and an average brain.”
— Thorsten Meyer, ThorstenMeyerAI.com
high performance AI language models
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Deployment Results May Differ
The supplied material does not establish how Super VL performs across independent production deployments or on specific customer workloads. The 630-token-per-second result came from one pre-release endpoint; provider-wide figures are lower and vary widely. It is also unclear whether all providers use the same hardware, serving configuration or measurement conditions.
The benchmark and pricing comparisons are reported by ThorstenMeyerAI.com using Artificial Analysis data, but the supplied material does not include full test protocols or independently verified invoices. Actual costs will depend on prompt length, caching, output volume and provider pricing. The claimed commercial-use permission is attributed to the stated license; teams still need to review the license terms for their own use.
Several task scores show improvement, but they do not settle whether the model’s reasoning delay or benchmark weaknesses will matter for a given application. The source reports an approximately 11-second wait before the first answer token in its reasoning profile, which could be acceptable for background processing but less suitable for interactive chat.
AI model licensing and licensing info
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Test Costs on Your Workload
The next practical test is a side-by-side evaluation using the provider and hardware a team intends to run. That should measure end-to-end latency, task accuracy, input and cache costs, and failure rates on representative prompts—not just output tokens per second or an aggregate index score.
Teams considering self-hosting will also need to verify hardware requirements and serving performance for the released NVFP4 weights. The source does not provide a complete deployment guide or establish the hardware configuration needed to reproduce its fastest result. Provider selection may materially change throughput and cost.
Artificial Analysis and providers may publish additional results as the release receives wider testing. Until then, the confirmed picture is a fast, commercially usable open-weight model with a 26-point index score and lower task cost than its predecessor, but no clear lead on intelligence value across the broader field.
As an affiliate, we earn on qualifying purchases.
Key Questions
What score did Nemotron 3.5 Super VL receive?
Artificial Analysis scored it at 26 on the Intelligence Index; the source’s specification table gives a more precise score of 26.5. It says the model ranks first among 67 open-weight models in the 40-billion-to-150-billion-parameter size class.
Is Nemotron 3.5 Super VL faster than other models?
A pre-release DeepInfra endpoint produced about 630 output tokens per second on 100,000-token prompts. Across six providers, the reported median was 307 tokens per second for prompts of that length, so results depend on the endpoint.
How much does an Intelligence Index task cost?
Artificial Analysis puts the cost at $0.25 per task. The source says input makes up about 89% of that cost, so real bills will depend on prompt size, caching and usage.
Can businesses use the model commercially?
The source identifies the license as OpenMDW-1.1 and says it permits commercial use. Businesses should review the license terms directly before deployment.
Does the score make it the best-value open model?
Not based on the comparisons in the source. It lists several open models with higher scores for similar or lower cost per benchmark task, while Nemotron’s case rests more on speed, license and the option to run it on NVIDIA hardware.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
