Key facts
| Model class | Fast long-context profile in the Plugsky catalogue |
| Best for | Document triage, RAG answering and large-prompt classification |
| Context class | 128K-class window; live limits are published per model |
| Capabilities | Streaming, function calling, JSON mode and long-context inputs |
| Pricing tier | Paid-plan model; free plan covers plugsky-micro and plugsky-lite |
| API | OpenAI-compatible /v1/chat/completions — no SDK changes |
| Resilience | Backup upstream plus same-profile fallback peers |
| Product status | Live |
TL;DR
- Long-context capacity with a speed-first design point.
- Best for document triage, RAG answering and big-prompt classification.
- Escalate to a stronger long-context tier when synthesis gets hard.
- Function calling and JSON mode keep it usable in pipelines.
- Read /models for the live window and current routing.
How it works, step by step
- Check the live model card at /models for the window, output limit and feature flags.
- Pick inputs where length is the challenge, not reasoning depth.
- Benchmark plugsky-nano against the free tiers on your real prompt lengths.
- Add retrieval or summarisation if the corpus outgrows a single call.
- Route complex synthesis to a stronger long-context tier and cap retries.
- Watch latency and quality percentiles, not averages, as prompt size grows.
Try it yourself
Open the context window comparison →
What plugsky-nano is
plugsky-nano is the fast long-context profile in the Plugsky catalogue. Where a top-tier model buys quality with latency, nano is positioned the other way around: it is built to keep long prompts moving. It keeps the standard OpenAI-compatible surface — streaming, function calling and JSON mode — and is listed as 128K-class on the live model card.
Profiles are served by an upstream engine that can change, so /models remains the source of truth for the live window, feature flags and routing.
When to choose it
Choose plugsky-nano when input length is the problem and the reasoning is not especially deep. Fitting examples include triaging long tickets, answering questions over a single large document, classifying large prompts and summarising retrieved passages before a smaller answer step.
It also works as the middle rung of a long-context ladder: try nano first, escalate to plugsky-longctx or plugsky-kimi when the answer needs multi-document synthesis, and move to a retrieval design when the corpus keeps growing.
Long-context trade-offs
Every long-context call pays for the whole input, and quality tends to drop in the middle of very long prompts. Structure beats size:
- Put instructions and the question at the edges, not buried mid-prompt.
- Trim boilerplate before sending; retrieval is often cheaper than a giant prompt.
- Keep output limits deliberate so a long input does not crowd out the answer.
- Re-test when the model card changes — long-context quality is sensitive to upstream swaps.
Self-serve plans are flat monthly with fair-use usage, so see the live pricing page for how plans map to sustained long-input traffic.
How to switch to plugsky-nano
Switching is one model name on the same endpoint: {"model": "plugsky-nano", "messages": [{"role": "user", "content": "Summarise the key obligations in this contract."}]}.
Test at your real prompt lengths rather than with short samples, because long-input behaviour is where the model earns its place. Keep a stronger long-context tier configured for escalations so quality gaps never reach users.
Honest comparison
| Dimension | plugsky-nano | plugsky-longctx | Free models (micro, lite) |
|---|---|---|---|
| Best fit | Long inputs with a latency budget | Long inputs needing more synthesis | Short, high-volume tasks |
| Speed | Fast tier | Moderate | Fast tier |
| Quality ceiling | Good for triage and extraction | Higher for analysis | Good for simple tasks |
| Context class | 128K-class | 128K-class | 128K-class |
| Use as | First long-context hop | Escalation for analysis | Zero-cost baseline |
| Plan | Paid | Paid | Free, no card |
Frequently asked questions
Is plugsky-nano free?
No — it is a paid-plan model. The free plan covers plugsky-micro and plugsky-lite with no card required, and a 14-day full-access trial lets you evaluate paid tiers first.
What is plugsky-nano best at?
Long inputs that must stay responsive: document triage, retrieval-augmented answering, large-prompt classification and summarising retrieved passages before a final answer.
How is it different from plugsky-longctx?
Both handle long inputs. plugsky-nano leans into speed, while plugsky-longctx targets longer, more analytical work. Benchmark both at your real prompt lengths and route by task.
What context window does it have?
It is 128K-class today, but the live window and output limit are published per model on the catalogue. Use that page as the source of truth.
Does it support tools and structured output?
Yes — streaming, function calling and JSON mode are part of the OpenAI-compatible surface, so tool-using pipelines work without changes.
Should I send my whole corpus to it?
No. Long prompts are expensive and quality can drift in the middle. Retrieve the relevant passages, put instructions at the edges, and reserve full-document sends for cases that truly need them.
How is pricing structured?
Self-serve plans are flat monthly with fair-use usage rather than per-token billing. See the live pricing page for current plans.