Key facts
| API compatibility | OpenAI-compatible /v1/chat/completions (drop-in base URL change) |
| Models | 30+ curated models with stable names you can pin |
| Pricing | Flat monthly self-serve plans with unlimited fair-use usage |
| Free tier | plugsky-micro and plugsky-lite, no card required |
| Trial | 14-day full-access trial for stronger models |
| Deployment | Plugsky cloud, your VPC, on-prem and air-gapped |
| Streaming and tools | Streaming, JSON mode and function calling are live |
TL;DR
- Routing layers add flexibility and operational surface.
- A curated provider reduces variables to model, prompt and price.
- OpenAI compatibility keeps the migration to configuration changes.
- Flat monthly pricing removes credit-balance management.
- Test model-for-model; similar names do not guarantee similar behaviour.
How it works, step by step
- Export the model list your application requests and their traffic shares.
- Map each model to the closest Plugsky tier and record expected differences.
- Create a Plugsky key on the free plan and run a replay against staging.
- Compare outputs, latency and failure modes per model.
- Migrate the highest-volume paths first, then the long tail.
- Keep the router configuration available until evals are stable.
Original data
Try it yourself
Open the OpenRouter API cost calculator →
What you get from a routing API
Routing APIs shine when availability and model variety matter more than operational simplicity. They let you switch upstreams, set fallbacks and test new models without changing your application. The cost is indirection: the same model name can behave differently depending on the host, and billing is spread across credits and upstream rates.
That indirection is manageable for exploration and harder to justify once production patterns are stable. At that point, most teams want fewer variables between their prompts and their users.
What changes with a curated provider
Plugsky provides one endpoint, 30+ models and a single operational contract. Model names are stable, streaming and tool calling are live, and embeddings share the same key, which simplifies RAG and agent stacks.
- One invoice and one set of limits.
- Flat monthly self-serve pricing instead of credit arithmetic.
- Deployment choices: cloud, VPC, on-prem or air-gapped.
- Free plan with plugsky-micro and plugsky-lite, plus a 14-day full-access trial.
Migration mechanics and evaluation
The mechanical change is small: base URL, key and model names. The substantive work is evaluation, because a curated model may be better or worse than the routed one you used for a specific task. Build a per-task rubric and measure correctness, format and latency.
Keep a rollback configuration so any single task can return to the router if quality drops. Track per-model error rates so a single curated model cannot quietly degrade the whole product. Audio, images, moderation, files, batch, fine-tuning, assistants and responses are coming soon; chat, tools, embeddings, RAG and agents are live today. See the live pricing page for current plans.
Honest comparison
| Capability | Plugsky | OpenRouter API | Direct multi-provider setup |
|---|---|---|---|
| Interface | OpenAI-compatible | OpenAI-compatible router | One client per provider |
| Catalogue | 30+ curated models | Very broad marketplace | Whatever you integrate |
| Pricing shape | Flat monthly self-serve | Credits and usage-based | Mixed bills |
| Deployment | Cloud, VPC, on-prem, air-gapped | Managed gateway | Varies by provider |
| Operational surface | One contract | Many upstreams | Many integrations |
Frequently asked questions
Can I migrate from OpenRouter without changing my app?
If you call the OpenAI-compatible chat completions endpoint, changing the base URL, key and model names is usually enough.
Will the same model name behave identically?
Not guaranteed. Treat each move as a model change: evaluate on your prompts and watch for differences in tone, formatting and tool-call behaviour.
Is there a free plan?
Yes. plugsky-micro and plugsky-lite are free with no credit card, and a 14-day full-access trial opens stronger models.
How is pricing structured?
Self-serve plans are flat monthly with unlimited fair-use usage and no per-token billing. See the live pricing page for current plans.
Can I use OpenRouter and Plugsky together?
Yes. Some teams keep the router for experiments and run production traffic on one curated provider, or add Plugsky as an upstream in the router.
Does Plugsky support embeddings and agents?
Yes. Embeddings, RAG and agents are live alongside chat, streaming and function calling.