Nvidia lists 20 models and Scaledown 2. Scaledown is cheaper on a typical model ($0.03 per 1M tokens median). Below: line-up, price tiers, the cheapest model for each capability, and what each costs for five real workloads.
| Nvidia | Scaledown | |
|---|---|---|
| Models listed | 20 | 2 |
| Models with published pricing | 20 | 2 |
| Typical price (median, 3:1 blend per 1M) | $0.19 | $0.03 |
| Cheapest paid model | NVIDIA: Switchyard ($-1000000.0000) | ScaleDown: Extract ($0.03) |
| Longest context | NVIDIA: Nemotron 3 Ultra (free) (1M) | ScaleDown: Extract (1M) |
| Free models | 8 | 0 |
| Newest model | NVIDIA: Switchyard (2026-09-21) | ScaleDown: Extract (2026-10-13) |
| Price changes in the last 90 days | 34 | 0 |
For each workload: the cheapest model from each provider that can actually do it (fits the context window; tool calling required for the agent), and its monthly cost at list price.
| Workload (per month) | Nvidia | Scaledown |
|---|---|---|
| Customer-support chatbot 50,000 conversations — each a 6-turn conversation | $-95000000.0000 with NVIDIA: Switchyard | $3.00 with ScaleDown: Extract |
| RAG search over documents 20,000 questions — each a question answered from ~8 retrieved passages | $-130000000.0000 with NVIDIA: Switchyard | $4.80 with ScaleDown: Extract |
| Coding / tool-using agent 2,000 tasks — each a multi-step task with tool calls and file context | $6.72 with NVIDIA: Nemotron Nano 9B V2 | no suitable model |
| Batch summarisation 10,000 documents — each a 6,000-word report summarised to a page | $-86000000.0000 with NVIDIA: Switchyard | $3.20 with ScaleDown: Extract |
| Long-document analysis 500 documents — each a 150-page contract or codebase read in one pass | $-51000000.0000 with NVIDIA: Switchyard | $2.00 with ScaleDown: Extract |
| You need | Nvidia | Scaledown |
|---|---|---|
| Any model | NVIDIA: Switchyard ($-1000000.0000) | ScaleDown: Extract ($0.03) |
| With tool calling | NVIDIA: Nemotron Nano 9B V2 ($0.07) | none listed |
| With image input | NVIDIA: Nemotron 3.5 Content Safety ($0.20) | ScaleDown: Extract ($0.03) |
| With reasoning mode | NVIDIA: Nemotron Nano 9B V2 ($0.07) | none listed |
| 128K+ context | NVIDIA: Switchyard ($-1000000.0000) | ScaleDown: Extract ($0.03) |
| 1M+ context | NVIDIA: Switchyard ($-1000000.0000) | ScaleDown: Extract ($0.03) |
| Model type | Nvidia | Scaledown |
|---|---|---|
| text+image+audio+video->text | 1 | 0 |
| text+image+file->text | 0 | 2 |
| text+image+video->text | 2 | 0 |
| text+image->text | 2 | 0 |
| text->text | 15 | 0 |
| Price per 1M (3:1 blend) | Nvidia | Scaledown |
|---|---|---|
| Free | 8 | 0 |
| Under $0.50 | 7 | 2 |
| $0.50 – $3 | 4 | 0 |
| $3 – $10 | 0 | 0 |
| $10 and up | 0 | 0 |
Scaledown is cheaper on a typical model: a median of $0.03 per 1M tokens (3:1 input/output) against $0.19 for Nvidia. The cheapest paid models are NVIDIA: Switchyard at $-1000000.0000 and ScaleDown: Extract at $0.03.
It depends on the job: Nvidia offers the broader line-up (20 models vs 2), Scaledown the lower typical price. The workload table above shows which wins for chat, RAG, agents, batch jobs and long documents.
Nvidia: Nvidia lists 20 models and Scaledown lists 2.
It is a tie: Nvidia's longest is NVIDIA: Nemotron 3 Ultra (free) at 1M tokens; Scaledown's is ScaleDown: Extract at 1M.
Nvidia: 8 free models (e.g. NVIDIA: Nemotron 3 Nano 30B A3B (free)). Scaledown: none.
Nvidia: 5 of 20 models accept images. Scaledown: 2 of 2.