Models

What is plugsky-micro and when should you use it?

plugsky-micro is one of the two free models on Plugsky's free plan, in a catalogue of 30+ models, alongside plugsky-lite. It is the free option for tasks that benefit from more reasoning: multi-step chat, tool use, structured extraction and analysis, with streaming, function calling and JSON mode on the OpenAI-compatible API. Start free with no card; use the 14-day full-access trial when you need paid tiers.

Key facts

Model classFree-plan reasoning-capable model in the Plugsky catalogue
Best forFree-agent chat, tool use, structured extraction and analysis
Free planOne of the two free models, together with plugsky-lite; no card required
Context class128K-class window; live limits are published per model
CapabilitiesStreaming, JSON mode and function calling
APIOpenAI-compatible /v1/chat/completions — keep your SDK
ResilienceAutomatic backup upstream plus same-profile fallback peers
Product statusLive

TL;DR

  • A free model for tasks that need more than the fastest possible answer.
  • Free plan access with no card, plus tools, JSON mode and streaming.
  • Compare it with plugsky-lite on your prompts — both are free.
  • Move to paid tiers only when your evaluation shows a real quality gap.
  • Automatic fallback keeps free-tier integrations stable.

How it works, step by step

  1. Create an account and generate an API key on the free plan — no card needed.
  2. Read the live model card at /models for the current window and feature flags.
  3. Call plugsky-micro with a real tool-calling or analysis prompt.
  4. Run the same prompts through plugsky-lite and compare quality and latency.
  5. Pick the free model that wins for each task in your application.
  6. Escalate tasks that fail your quality bar to a paid model when you are ready.
  7. Monitor usage and status through your dashboard and the status page.
1Create an accountand generate an APIkey on the free2Read the live modelcard at /models forthe current window3Call plugsky-microwith a realtool-calling or4Run the sameprompts throughplugsky-lite and5Pick the free modelthat wins for eachtask in your6Escalate tasks thatfail your qualitybar to a paid model

Try it yourself

Open the LLM cost calculator →

What plugsky-micro is

plugsky-micro is one of the two free models on Plugsky's free plan. Where lite optimises for speed on simple tasks, micro is the free option for requests that benefit from reasoning: multi-step chat, tool selection, structured extraction and light analysis. It exposes the same OpenAI-compatible surface as paid models, including streaming, function calling and JSON mode.

Free-tier routing is updated over time, so check the live card at /models for the engine, context window and capability flags currently in effect. Its free sibling is plugsky-lite.

When to choose it

Choose plugsky-micro when you want to build or validate a reasoning-capable feature at zero cost: an agent that calls a couple of tools, a form extractor that returns JSON, a classifier that must follow a short chain of logic, or a development environment for prompts before you move them to production models.

Move up to a paid model when accuracy, long-form synthesis or hard reasoning becomes business-critical. The 14-day full-access trial is the structured way to compare a paid tier against micro on your own workload before upgrading.

Free-tier trade-offs

Free models are excellent value and have real limits. The engineering discipline is to know which requests they pass and to route the rest.

  • Keep prompts short and explicit; small models reward clarity over volume.
  • Use JSON mode with schema validation for structured output.
  • Route low-confidence or failed results to a paid model rather than retrying the same prompt.
  • Track quality on a small labelled set so upgrades are evidence-based.
  • Compare micro and lite before assuming which free model suits a feature.

How to switch to plugsky-micro

Switching is a base URL plus a model name: {"model": "plugsky-micro", "messages": [{"role": "user", "content": "Extract the order details as JSON."}], "response_format": {"type": "json_object"}} posted to https://api.plugsky.com/v1/chat/completions.

Existing OpenAI-compatible code works unchanged. When a task outgrows the free tier, change one model name to a paid model and keep the rest of the integration intact.

Honest comparison

Dimensionplugsky-microplugsky-litePaid reasoning model
PlanFree, no card requiredFree, no card requiredPaid plan
Best fitFree tasks needing reasoning and toolsFast simple chat at high volumeProduction-grade deep work
Latency profileFastFastest tierSlower by design
Tools and JSON modeSupportedSupportedSupported
Context class128K-class128K-class128K-class
FailoverAutomatic peer fallbackAutomatic peer fallbackAutomatic peer fallback

Frequently asked questions

Is plugsky-micro free?

Yes. The free plan includes two free AI models, plugsky-micro and plugsky-lite, and no credit card is required to start.

How is it different from plugsky-lite?

Both are free. Micro is the reasoning-capable free option; lite is positioned for the fastest simple responses. Test both on your own prompts to see which fits each task.

What context window does it have?

It is 128K-class today. The exact live window and output limit are published per model at /models.

Can it call tools?

Yes — function calling, JSON mode and streaming are part of the shared OpenAI-compatible surface.

Is it good enough for production?

For tasks it passes on your evaluation, yes. Keep an escalation path to a paid model for requests that need deeper reasoning or higher accuracy.

How does the paid trial work?

A 14-day full-access trial lets you test paid models on your workload so you can decide with evidence rather than guesswork.

What happens if the free model has an incident?

Requests retry through a backup upstream and same-profile fallback peers. Live component health is on the status page.

Can I use my existing OpenAI SDK?

Yes. Change the base URL to the Plugsky endpoint and set the model name; your prompts, tools and response parsing stay the same.