NVIDIA
NVIDIA ranks #20 of 91 providers on BenchLeader by its best model, with 16 ranked models. Its best model is Nemotron 3 Ultra 550B A55B at 55.7 on the BenchLeader Index, strongest in instruction following. Prices run from $0.070 to $1.13 per million tokens blended; NVIDIA Nemotron Nano 9B v2 is the cheapest ranked option. The fastest is Nemotron 3.5 Lightning at 288 tokens per second. 20 of its 39 models ship open weights. It has shipped 17 dated releases, roughly one every 36 days; the last was 23 Oct 2026, so the next would be due between 13 Nov 2026 and 12 Dec 2026 on that pattern.
The GPU maker also publishes open Nemotron models fine-tuned from Llama and its own architectures, mostly as showcases for its inference stack.
- Provider rank
- #20
- by best model
- Best index
- 55.7
- Nemotron 3 Ultra 550B A55B (thinking)
- Ranked models
- 16
- 20 open weights
- Cheapest ranked
- $0.070/M
- NVIDIA Nemotron Nano 9B v2
Release timeline
Each model’s best configuration by release date. The top eight are labelled; hover any point for its name.
- NVIDIA16
Next release
Over the last two years NVIDIA shipped 17 distinct models, one every 36 days on median. The latest arrived on 23 Oct 2026.
If the cadence holds, the next one lands between 13 Nov 2026 and 12 Dec 2026.
An estimate from past release dates only. It knows nothing about roadmaps or announcements.
Fastest ranked model: Nemotron 3.5 Lightning at 288 tok/s.
Head to head
NVIDIA vs Anthropic
NVIDIA vs Google
NVIDIA vs Meta
NVIDIA vs OpenAI
NVIDIA vs Moonshot AI
NVIDIA vs SpaceXAI
NVIDIA vs DeepSeek
NVIDIA vs Xiaomi
NVIDIA vs Zhipu AI
NVIDIA vs Alibaba
NVIDIA vs MiniMax
NVIDIA vs StepFun
NVIDIA vs Sapiens AI
NVIDIA vs ByteDance
NVIDIA vs Motif Technologies
NVIDIA vs Mistral AI
NVIDIA vs MBZUAI
NVIDIA vs Meituan
NVIDIA vs Multiverse ComputingModels
Every NVIDIA model with a result, one row per model; switch to all configurations to see each reasoning effort.
| # | ||||||||
|---|---|---|---|---|---|---|---|---|
| 1 | Nemotron 3 Ultra 550B A55Bthinkingopen ↗ | 55.7±11 | $1.13 | 154 | 16 s | 1M | 2026-06-04 | 2026-06-04 |
| 2 | Nemotron 3 Super 120B A12Bthinkingopen ↗ | 50.6±8 | $0.450 | 155 | 14 s | 262k | 2026-03-11 | 2026-03-11 · stale |
| 3 | Nemotron 3.5 Super VL 120B A12Bopen | 50.4±9 | $0.788 | – | – | 1.0M | 2026-10-15 | 2026-10-23 |
| 4 | Nemotron Cascade 2 30B A3Bopen ↗ | 49.6±11 | – | – | – | 1M | 2026-03-19 | 2026-03-19 · stale |
| 5 | Llama Nemotron Ultra 253Bopen | 48.1±10 | – | – | – | 128k | 2025-04-07 | 2026-10-22 |
| 6 | Nemotron 3.5 Lightningopen ↗ | 47.4±6 | $0.095 | 288 | 7.59 s | 1M | 2026-08-11 | 2026-08-11 |
| 7 | Nemotron 3 Nano 30B A3Bthinkingopen ↗ | 46.5±8 | $0.088 | 235 | 9.46 s | 262k | 2025-12-15 | 2025-12-15 · stale |
| 8 | Nemotron 3 Nano Omni 30B A3Bthinkingopen | 45.1±6 | $0.643 | 234 | 9.33 s | 256k | 2026-04-29 | 2026-04-29 |
| 9 | Nemotron 3 Super | 44.3±3 | – | – | – | – | 2026-03-11 | 2026-03-11 · stale |
| 10 | Nemotron 3 Ultra | 43.9±11 | – | – | – | – | 2026-06-04 | 2026-10-23 |
| 11 | Llama Nemotron Super 49B v1.5thinkingopen | 43.6±1 | – | – | – | 128k | 2025-07-25 | 2025-07-25 · stale |
| 12 | Nemotron 3 Nano 4Bopen | 43.1±7 | – | – | – | 262k | 2026-03-16 | 2026-03-16 · stale |
| 13 | Llama 3.1 Nemotron 70B Instructopen ↗ | 42.0±3 | – | – | – | 128k | 2024-10-15 | 2026-10-22 |
| 14 | Nemotron Nano 12B v2 VLthinkingopen | 41.5±3 | – | – | – | 128k | 2025-10-28 | 2025-10-28 · stale |
| 15 | Llama 3.3 Nemotron Super 49B v1.5thinkingopen | 41.5±3 | $0.400 | – | – | 131k | 2025-03-18 | 2025-03-18 · stale |
| 16 | NVIDIA Nemotron Nano 9B v2thinkingopen | 41.5±3 | $0.070 | 95 | 22 s | 131k | 2025-08-18 | 2025-08-18 · stale |
| 17 | Langston Nim Nvidia Llama 3.3 Nemotron Super 49B v1 | – | – | – | – | – | – | 2026-09-01 |
| 18 | Langston Nim Nvidia Llama 3.3 Nemotron Super 49B v1 42e84561thinking | – | – | – | – | – | – | 2026-09-01 |
| 19 | Llama 2 70B Steerlm | – | – | – | – | – | – | 2026-10-22 |
| 20 | Llama 3.1 Nemotron 51B | – | – | – | – | – | – | 2026-10-22 |
| 21 | Nemotron 3.5 Bolt 74B-A9Bopen | – | – | – | – | 262k | 2026-10-15 | 2026-10-15 |
| 22 | Nemotron 3.5 Lightning 30B A3Bopen | – | $0.262 | – | – | 262k | 2026-08-11 | 2026-10-22 |
| 23 | Nemotron 3.5 Super VLopen ↗ | – | $0.315 | 453 | 5.12 s | 1.0M | 2026-10-15 | 2026-10-15 |
| 24 | Nemotron 4 340B | – | – | – | – | – | – | 2026-10-22 |
| 25 | Nemotron Lightning 3.5 30B A3B | – | – | – | – | – | – | 2026-09-28 |
| 26 | NewNemotron 3.5 Bolt E4B 4GBthinking | – | – | – | – | 131k | 2026-10-23 | 2026-10-23 |
| 27 | Nemotron Lightning 3.5 | – | – | – | – | – | – | 2026-10-14 |
| 28 | Nemotron Super VL 3.5 | – | – | – | – | – | – | 2026-10-14 |
| 29 | NewNemotron Super VL 3.5 NVFP4 | – | – | – | – | – | – | 2026-10-14 |
| 30 | Nemotron Ultra 3 | – | – | – | – | – | – | 2026-10-14 |
| 31 | Nvila Internal 15B v1 | – | – | – | – | – | – | 2026-10-19 |
| 32 | NewSmaug Nemotron 3.5 NVFP4 | – | – | – | – | – | – | 2026-10-14 |
| 33 | Cosmos Reason2 8Bopen ↗ | – | – | – | – | 131k | 2025-12-01 | – |
| 34 | Megatron-Turing NLG 530B | – | – | – | – | – | 2021-10-11 | – |
| 35 | Nemotron 3 Ultra Nvfp4high | – | – | – | – | – | – | – |
| 36 | Nemotron-4 15B | – | – | – | – | – | 2024-02-26 | – |
| 37 | nemotron-voicechatopen | – | – | – | – | 128k | 2026-03-16 | – |
| 38 | NVILA 8B | – | – | – | – | – | 2024-12-10 | – |
| 39 | Riva Translate 4B Instruct v1.1open | – | – | – | – | 128k | 2025-12-12 | – |
Latest news
- Running Frontier-Class AI Models Across 8 NVIDIA DGX Spark Nodes | NVIDIA Technical BlogNVIDIA Developer · 22 Oct 2026
- Nvidia launches new open models to run agents in the enterprise and on PCsTechCrunch · 19 Oct 2026
- Nvidia and CrowdStrike Develop New Cybersecurity AI Models | The Morning Download for Sept. 2WSJ · 2 Sept 2026
- Code, Test, and Deploy on One Machine with NVIDIA RTX Spark | NVIDIA Technical BlogNVIDIA Developer · 16 Oct 2026
- NVIDIA Nemotron 3.5 Super VL Powers Specialized Agents with Multimodal Reasoning | NVIDIA Technical BlogNVIDIA Developer · 15 Oct 2026
- Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt ModelsNVIDIA Blog · 15 Oct 2026
- Nvidia's role as AI financier faces first big test as neoclouds take market hitCNBC · 24 Oct 2026
- UW protein institute led by Nobel winner unveils its first AI model co-developed with NvidiaGeekWire · 21 Oct 2026