Nemotron 3.5 Super VL
NVIDIAReleased Oct 15, 2026Open weights · OpenMDW-1.1nemotron-3.5-super-vl
- Reasoning
- yes
- Context
- 1M tokens
- Max output
- 128k tokens
- Input
- text, image
- API features
- tools, structured outputs, JSON mode, seed, logprobs
Read Oct 23, 06:00 UTC from the hosts' published offers and OpenRouter, spec sheet from NVIDIA
Free right now
- Free OpenRouter, served by Nvidia · no end date · 20 req/min, 50 req/day, 1,000 req/day with $10 of credits
Checked Oct 23 · every free model on Deals
The cheapest host, Crusoe, serves a 256K context (maker: 1M). Weights and context per host
Speed by provider
Output speed of Nemotron 3.5 Super VL at each host, fastest first
- CoreWeave388 tok/s0.93 s to first token
- Crusoe309 tok/s0.39 s to first token
- Together AI301 tok/s0.34 s to first token
- BaseTen287 tok/s0.63 s to first token
- Io Net241 tok/s0.57 s to first token
- DeepInfra44 tok/s1.06 s to first token
- Venice44 tok/s0.80 s to first token
Speed: OpenRouter, median of the last 30 min, read Oct 23, 06:00 UTC · a snapshot of real traffic before the daily read, not a daily average · bars from zero.
Run it yourself
open weights, OpenMDW-1.1 · fits 4 of the 6 workstations in our guideSelf-host guide ›- Parameters
- 120B
- 12B active per token (MoE)
- Memory needed
- 72 GB
- in INT4 · 144 GB in INT8
- Smallest cloud setup
- A100
- 80 GB · $1.19/h, about $869 a month
- Worth it above
- 2B tokens
- a month, against the cheapest API at $0.50 per M
| Quantization | Memory | Smallest cloud GPU | Quality |
|---|---|---|---|
| INT4 (Q4) | 72 GB | A100 · $1.19/h | smallest footprint, minor quality loss |
| INT8 (Q8) | 144 GB | 2× H100 · $4.30/h | near-lossless |
| FP16 (full) | 288 GB | 4× H100 · $8.60/h | reference quality |
At home: in INT4 it fits a Mac Studio M5 Ultra (96 GB unified), about $5,499 once, or $0.2092/h over 3 years, electricity aside; it pays for itself after about 11B tokens against the cheapest API.
Memory ≈ parameters × bytes per parameter × 1.2 · cloud price: on-demand rate, running 24/7 · builds: GGUF, AWQ, MLX, GPTQ, Ollama · break-even calculator
| Provider | Input | Output | Cache ↑ | tok/s | First token | Uptime 24h | Checked |
|---|---|---|---|---|---|---|---|
| Together AInvfp4 (bf16) · 256K ctx (1M) | $0.15 | $0.50 | $0.02 | 301 | 0.34 s | 99.9% | Oct 23 |
| DeepInfrafp4 (bf16) · 256K ctx (1M) | $0.15 | $0.50 | $0.02 | 44 | 1.06 s | 97.2% | Oct 23 |
| CoreWeavenvfp4 (bf16) · 1M ctx | $0.15 | $0.50 | $0.02 | 388 | 0.93 s | 97.5% | Oct 23 |
| Io Netnvfp4 (bf16) · 256K ctx (1M) · 128K out | $0.15 | $0.50 | $0.02 | 241 | 0.57 s | 95.7% | Oct 23 |
| BaseTenfp4 (bf16) · 1M ctx | $0.20 | $0.50 | $0.04 | 287 | 0.63 s | 100% | Oct 23 |
| Crusoebf16 · 256K ctx (1M) | $0.10 | $0.40 | $0.08 | 309 | 0.39 s | 95.6% | Oct 23 |
| Venicefp4 (bf16) · 256K ctx (1M) · 32K out | $0.25 | $1.25 | $0.13 | 44 | 0.80 s | 92.8% | Oct 23 |
7 of 7 providers · prices per million tokens, read Oct 23, 06:00 UTC · speed: OpenRouter, median of the last 30 min, read Oct 23, 06:00 UTC · uptime: OpenRouter, last 24 h · under each host: weights precision, context and max output as served, the maker's value in brackets when the host's differs
| Provider | Weights | Context | Max output | Tools | Structured outputs | Reasoning |
|---|---|---|---|---|---|---|
| Crusoe | bf16 | 256K | not stated by the host | yes | yes | yes |
| Together AI | nvfp4 | 256K | not stated by the host | yes | yes | yes |
| DeepInfra | fp4 | 256K | not stated by the host | yes | yes | yes |
| CoreWeave | nvfp4 | 1M | not checked yet | yes | yes | yes |
| Io Net | nvfp4 | 256K | 128K | yes | no | yes |
| BaseTen | fp4 | 1M | not checked yet | yes | no | yes |
| Venice | fp4 | 256K | 32K | yes | yes | yes |
Weights, context and max output: read at each host's own catalogue or docs where its terms allow, otherwise as the host declares to OpenRouter; hover a cell for its source and date · tools, structured outputs, reasoning: as declared to OpenRouter · not verified by us
Nemotron 3.5 Super VL spec sheet, field by field, with sources
Nemotron 3.5 Super VL spec sheetmaker-declared
What NVIDIA declares about the model itself — read from its own weights repository and documentation, not from any host. What each endpoint actually serves is in the pricing above: the two can differ, and both are true.
- Context window
- 1,048,576 tokens
- Parameters
- 124,453,395,976≈ 124.5B
- Active experts
- 22 of 512 active per token
- Weight precision
- BF16
- License
- OpenMDW-1.1 — open weights, permissiveScope: the licence on the published weights, and nothing else. It is not the maker's acceptable-use policy, which is a separate document, and it is not the contract you sign with whichever host you call — hosts set their own terms, which this index does not read.
- Knowledge cutoff
- not published by the maker
- Modalities
- text · image
- Max output (official)
- not published by the maker
- Release date
- October 15, 2026
- Maker lifecycle status
- not published by the maker
Also declared in the weights
- Architecture class
- NemotronH_Omni_Reasoning_V3
- Vocabulary
- 131,072 tokens
- Weights first published
- October 12, 2026
- Weights last updated
- October 15, 2026
- Languages declared
- English · Spanish · French · German · Italian · Japanese
nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 · config.json @ a1ccf94 · read 2026-10-15 · 7 of 10 fields published by the maker
| Date | Provider | Input | Output | Change |
|---|---|---|---|---|
| Oct 20 | DeepInfra | $0.10 → $0.15 | $0.40 → $0.50 | +30% |
| Oct 17 | CoreWeave | $0.15 → $0.15 | $0.75 → $0.50 | −28% |
| Oct 17 | DeepInfra | $0.20 → $0.10 | $1.00 → $0.40 | −58% |
Every published change of input or output price, newest first · change on input + output · full series: pricing-history.csv.
| Provider | Chatbot 1,000 in / 500 out | RAG / long context 8,000 in / 500 out | Batch summarize 4,000 in / 1,000 out |
|---|---|---|---|
| Crusoe | $0.30 | $1.00 | $0.80 |
| Together AI | $0.40 | $1.45 | $1.10 |
| DeepInfra | $0.40 | $1.45 | $1.10 |
| CoreWeave | $0.40 | $1.45 | $1.10 |
| Io Net | $0.40 | $1.45 | $1.10 |
| BaseTen | $0.45 | $1.85 | $1.30 |
| Venice | $0.88 | $2.62 | $2.25 |
What each host bills for 1,000 requests of a representative job
Your own workload
Other NVIDIA models · all its models, hosts and history ›
- Nemotron 3 Nano 30B A3B4 providers · from $0.25
- Nemotron 3 Super4 providers · from $0.48
- Nemotron 3.5 Lightning4 providers · from $0.18
- Nemotron 3 Ultra3 providers · from $2.70
- Nemotron 3.5 Content Safety1 provider · from $0.40
- Nemotron 3 Nano Omni 30B A3B Reasoning0 provider · from —