Nemotron 3.5 Lightning on QuickSilver Pro
Nemotron 3.5 Lightning is NVIDIA's open 30B-parameter Mixture-of-Experts model with only 3B active parameters per token. It is built for fast, tool-heavy agents, high-throughput automation, and domain customization. QuickSilver Pro provides 262K-token context at $0.08 input / $0.20 output per million tokens, with zero data retention and no markup.
At a glance
Fast, inexpensive tool-using agents — 30B total parameters, 3B active per token, and 262K context.
Pricing comparison ($/1M tokens)
| Provider | Input | Output | vs QSP |
|---|---|---|---|
| QuickSilver Pro | $0.08 | $0.20 | — |
| Market price (public paid rate) | $0.08 | $0.20 | same |
| OpenAI (GPT-4o-mini) | $0.15 | $0.60 | 67% lower |
When to use
Use Nemotron 3.5 Lightning for high-volume coding assistants, retrieval agents, long-running tool loops, structured extraction, and workflows that repeatedly reuse a large system prompt or repository context. Its sparse 30B-A3B design keeps generation inexpensive, and cached input costs $0.04 per million tokens.
When to use something else
This is a fast efficiency model, not a frontier flagship. Escalate difficult research, mathematical reasoning, or complex autonomous coding to DeepSeek V4 Pro, Qwen3.8 Max, or Grok 4.6 when your evaluations justify the higher price. The QSP endpoint is text-only and currently guarantees 262K-token context with zero data retention.
Quickstart (curl)
curl https://api.quicksilverpro.io/v1/chat/completions \
-H "Authorization: Bearer $QSP_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "nemotron-3.5-lightning",
"messages": [{"role": "user", "content": "Hello!"}]
}'OpenAI-compatible. One-line migration via base_url.
FAQ
$0.08 per million input tokens, $0.20 per million output tokens, and $0.04 per million cached-input tokens. QuickSilver Pro matches the lowest stable paid market rate without a markup.
The model architecture can support up to 1M tokens, while QSP currently guarantees 262K tokens in production. We publish the limit we test and enforce rather than advertising a theoretical maximum.
Yes. It supports selectable reasoning, streaming, function calling, and JSON Schema structured output. QSP defaults reasoning off for direct low-cost responses; pass `reasoning: { enabled: true }` when a task benefits from a reasoning trace.
Set base_url=https://api.quicksilverpro.io/v1, use your QSP key, and set model="nemotron-3.5-lightning". Both Chat Completions and Responses API clients use the same public model ID.