🔍 Read the full analysis: Nemotron 3.5 Super VL’s Three-Part Pitch: Cost, Speed, Score on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- 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 a score of 26, while reporting low benchmark task costs and high speed on one pre-release endpoint; results vary across providers, and several alternatives score higher for similar or lower cost.
Artificial Analysis scored the model at 26.5 on its Intelligence Index, rounded to 26 in the headline comparison. That is twice the score of its predecessor, Nemotron 3 Super, which scored 13. The new model ranked first among 67 open-weight models in the 40-billion-to-150-billion total-parameter size class, according to the supplied benchmark material. Across all open-weight models, it ranked about 18th. It has 120 billion total parameters and 12 billion active parameters in a mixture-of-experts design, and supports a context window of up to one million tokens.
The speed headline comes from a pre-release DeepInfra endpoint serving NVFP4, or four-bit, weights. Artificial Analysis measured about 630 output tokens per second with prompts of 100,000 tokens, describing it as the fastest model it had measured at that prompt length among models scoring at least 20. The model page’s median across six providers was lower: 196 tokens per second on its standard prompt and 307 with 100,000-token prompts. Provider results ranged from 106 to 534 tokens per second, so the single-endpoint result does not describe every deployment.
Artificial Analysis listed API prices of $0.20 per million input tokens and $0.575 per million output tokens, with cached input at $0.055 per million. It estimated $0.25 for a full Intelligence Index task, compared with $1.64 for Nemotron 3 Super. The launch write-up attributed the lower task cost in part to lower output volume: about 46,000 tokens per task versus 114,000 for the predecessor. The supplied cost breakdown says roughly 89% of the new model’s task cost comes from input, including cache writes; output accounts for less than three cents.
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 Deployment Trade-offs
For teams running long prompts or background agents, low latency and high throughput can matter as much as a broad capability score. Artificial Analysis reported a median time to first token of 0.89 seconds. It also said the model’s reasoning phase takes about 11 seconds before the first answer token, with a 500-token answer taking about 13.6 seconds end to end. That profile may suit batch work and automated pipelines better than conversations where users expect an immediate response.
The benchmark price also depends on the workload. Since input accounts for most of the measured task cost, teams that repeatedly send large documents or code repositories should evaluate their own prompt and caching patterns. The per-task figure is an estimate from a benchmark, not a guaranteed bill for a production workload.
The comparison set puts the score in perspective. The source lists MiMo-V2.6-Pro at 46 for $0.13 per task, MiMo-V2.6-Flash at 38 for $0.06, and Inkling Small at 26 for $0.09. These are Artificial Analysis comparisons, not results from every real-world use case. Nemotron’s case may rest instead on its measured speed, NVIDIA deployment options, and licensing terms.
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 Launch Fits The Lineup
Nemotron 3.5 Super VL follows Nemotron 3 Super, whose Intelligence Index score was 13. The new model’s score doubles that result, while the benchmark write-up reports a drop in cost per index task from $1.64 to $0.25. Artificial Analysis also recorded a reduction in the predecessor’s AA-Omniscience hallucination rate, from 87% to 36%. That remains a benchmark result; it does not establish how often the model will hallucinate across other settings.
The release material highlights gains on selected evaluations. GDPval-AA, which measures economically useful work, reached 1,162 Elo, compared with 490 for Nemotron 3 Super. The source also reports 999 Elo on AA-Briefcase, versus 876 for Nemotron 3 Ultra, and 14.8% on AutomationBench-AA, compared with 3.0% for Ultra. Those figures identify areas of improvement, but the AutomationBench result remains low in absolute terms and does not establish broad reliability in software workflows.
The model’s license is OpenMDW-1.1, which the supplied material says allows commercial use. The release also offers NVFP4 weights and the option to run the model on NVIDIA hardware. License permissions and the ability to self-host may be central for some organizations, even when other models post higher scores.
As an affiliate, we earn on qualifying purchases.
What The Benchmarks Do Not Settle
The reported speed varies considerably by provider, and the 630-token-per-second figure came from one pre-release endpoint. Production performance remains provider- and setup-dependent; the supplied material does not report independent testing across hardware configurations or sustained production workloads. It also does not specify whether the listed API prices will remain fixed.
The benchmark comparisons do not establish which model will perform best on a particular company’s tasks. The source provides selected scores and prices, but not the full evaluation methods, confidence ranges, or task-by-task results needed to judge every comparison. It is also unclear from the supplied information how the one-million-token context window performs on long, complex inputs in practice.
The material describes the license as permitting commercial use, but organizations still need to review the license terms for their own deployment. The reported reduction in hallucinations is relative to Nemotron 3 Super on a named benchmark; it is not a claim of error-free output. The source offers no basis for extending that rate to all prompts or uses.
open-weight AI models for commercial use
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Test The Intended Workload
The next useful comparison is at the endpoint and workload a team plans to use: measure latency, throughput, input and output token use, and task quality on representative prompts. Provider-level results should guide deployment decisions, since the reported speed spread is wide and input costs dominated the benchmark task estimate.
Teams considering commercial use or self-hosting can review OpenMDW-1.1 and NVIDIA’s deployment requirements alongside those measurements. Further provider benchmarks, production evaluations, or updated pricing could change the cost and speed picture; the supplied material does not give a date for any such update.
As an affiliate, we earn on qualifying purchases.
Key Questions
When did NVIDIA release Nemotron 3.5 Super VL?
NVIDIA released Nemotron 3.5 Super VL on October 15, 2026. Artificial Analysis published benchmark figures the same day, according to the supplied source.
What score did it receive?
Artificial Analysis reported an Intelligence Index score of 26.5, commonly rounded to 26. It ranked first among open-weight models in its 40-billion-to-150-billion total-parameter size class, but about 18th among open-weight models overall.
How fast is the model?
A pre-release DeepInfra endpoint reached about 630 output tokens per second on prompts of 100,000 tokens. The median across six providers was 307 tokens per second on those prompts, with reported results ranging from 106 to 534.
How much does a benchmark task cost?
Artificial Analysis estimated about $0.25 per Intelligence Index task. That is a benchmark estimate; actual costs depend on token use, caching, provider, and workload. The listed API rates were $0.20 per million input tokens and $0.575 per million output tokens.
Can businesses use the model commercially?
The supplied release material identifies the license as OpenMDW-1.1 and says it permits commercial use. Organizations should review the license terms and deployment requirements for their intended use.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
