Key facts
| Model count | 30+ models behind one OpenAI-compatible API |
| Free tier | plugsky-micro and plugsky-lite on the free plan, no card |
| Coding | plugsky-coder |
| Reasoning and general | plugsky-deepseek-flash, plugsky-deepseek-pro, plugsky-max, plugsky-frontier |
| Open-weight families | Includes Llama 4, Gemma, Mistral, Kimi, MiniMax, GPT-OSS and DeepSeek-class models |
| Embeddings | plugsky-embed and plugsky-embed-multilingual |
| Catalogue source | The live /models page is authoritative and updated |
| Coming soon | Audio, images, moderation, files, batch, fine-tuning, assistants, responses endpoints |
TL;DR
- 30+ models, one API, one integration.
- Free plan covers plugsky-micro and plugsky-lite with no card.
- Coding, reasoning, long-context and embedding models are all live.
- Open-weight families sit alongside Plugsky's own model tiers.
- Always confirm against the live catalogue; model line-ups change.
How it works, step by step
- List your workloads and classify what each needs: speed, reasoning, context, embeddings.
- Shortlist two or three models per workload from the live catalogue.
- Benchmark candidates on your own evaluation set, not generic leaderboards.
- Pin chosen models or aliases in configuration rather than hard-coding them.
- Enable fallback to a second model for each critical path.
- Revisit the catalogue quarterly and re-run evals when models update.
Try it yourself
Open the model availability tracker →
How the catalogue is organised
Think in tiers rather than memorising names:
- Free tier:
plugsky-microandplugsky-lite, included on the free plan and suitable for prototyping, classification and light generation. - Everyday workhorses: fast, economical models including
plugsky-deepseek-flashandplugsky-coderfor volume and coding. - Reasoning tier:
plugsky-deepseek-pro,plugsky-maxandplugsky-frontierfor harder tasks. - Specialised:
plugsky-longctxfor large contexts andplugsky-embed/plugsky-embed-multilingualfor retrieval. - Open-weight families: models in the Llama 4, Gemma, Mistral, Kimi, MiniMax, GPT-OSS and DeepSeek classes.
Every model uses the same API, so switching is a configuration change. The model catalogue is the authoritative, current list.
Models versus endpoints
Two different questions get confused: which models are available, and which API surfaces exist. Plugsky's chat, streaming, JSON mode, function calling and embeddings endpoints are live across the current catalogue. Other surfaces — audio, images, moderation, files, batch, fine-tuning, assistants and responses — are coming soon and are not part of the live API today. So it is accurate to say Plugsky supports a broad model catalogue through the chat and embedding endpoints, while specialist endpoints are still on the roadmap. Check the docs for the endpoint matrix before designing around any single capability.
What we do and what we do not do
What we do: keep a broad catalogue behind one stable API, document model aliases and capabilities, and update the live catalogue as models are added or retired. What we do not do: promise that a model listed today will remain forever — line-ups change across the industry, and we would rather retire a model transparently than keep a stale name alive. Pin versions or aliases where stability matters, re-run evaluations periodically, and treat the catalogue page, not this article, as the source of truth.
Honest comparison
| Need | Plugsky | Single-vendor API | Self-hosted stack |
|---|---|---|---|
| Model breadth | 30+ models, one API | Vendor family only | Whatever you deploy |
| Free tier | plugsky-micro and plugsky-lite, no card | Varies | Hardware cost |
| Coding models | plugsky-coder and others | Vendor-dependent | Open-weight options |
| Embeddings | plugsky-embed family, live | Often separate service | You operate it |
| Endpoint coverage | Chat and embeddings live; more coming soon | Varies | You build everything |
| Operational effort | Low | Low | High |
Frequently asked questions
How many models does Plugsky support?
30+ models are available behind one OpenAI-compatible API, spanning free-tier chat models, coding, reasoning, long-context, open-weight families and embeddings.
Which models are free?
The free plan includes plugsky-micro and plugsky-lite with no card required. The 14-day full-access trial covers stronger models.
Which model should I pick for coding?
Start with plugsky-coder for everyday coding and escalate complex refactors to a stronger reasoning model. Validate on your own repository.
Are open-weight models available?
Yes — the catalogue includes open-weight families such as Llama 4, Gemma, Mistral, Kimi, MiniMax, GPT-OSS and DeepSeek-class models, all behind the same API.
Where is the authoritative model list?
The live model catalogue page. This article summarises the shape of the catalogue, but the site page is updated as models change.
Do all models support the same features?
No. Chat, streaming, JSON mode and function calling availability varies by model. Check the capability matrix before depending on a feature.
Are audio, image and fine-tuning models available?
Those endpoints are coming soon, not live. Today's API covers chat, streaming, JSON mode, function calling and embeddings.