Key facts
| Model class | Largest general model in the Plugsky model ladder |
| Best for | Complex multi-step reasoning, deep analysis and long-form planning |
| Context class | 128K-class window; live limits are published per model |
| Capabilities | Streaming, function calling, JSON mode and long-context |
| 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 top-tier fallback peers |
| Product status | Live |
TL;DR
- The heavyweight option for problems with many steps and high stakes.
- Best used as an escalation tier, not as your default model.
- Tools, JSON mode and long context keep it compatible with existing agents.
- Compare against plugsky-frontier on your own prompts before standardising.
- Automatic peer fallback protects top-tier availability.
How it works, step by step
- Identify the request types that consistently fail on your current model.
- Read the live card at /models for window, output limit and feature flags.
- Run those failing requests through plugsky-max and measure the difference.
- Compare with plugsky-frontier on the same set to choose the right top tier.
- Build an escalation rule — failures, complexity signals or premium tiers.
- Switch qualifying traffic by changing the model name, keeping a fast default.
- Review escalation volume regularly to keep cost aligned with value.
Try it yourself
Open the LLM cost calculator →
What plugsky-max is
plugsky-max is the largest general model in the Plugsky ladder. Size buys depth: more capacity to hold a problem's constraints, follow long chains of reasoning and produce carefully structured output. It is intended for the requests where a cheaper model's answer is not good enough — architecture decisions, complex planning, dense analytical work, and long-form documents that must stay coherent.
The live card at /models lists the current context window, output limit and feature flags. Because top-tier routing changes most often, verify there before you build constraints into an application.
When to choose it
Choose plugsky-max as the escalation tier for your most demanding requests, and as a comparison point when choosing between top-tier models. It is not a good default: every request pays for depth it may not need.
For routine work use a balanced model such as plugsky-pro; for coding, the coder family; for long documents, plugsky-longctx. Benchmark plugsky-max against plugsky-frontier on your hardest prompts and route to whichever wins for each workload.
Trade-offs of the largest model
Bigger models are slower and more expensive per request. The right architecture confines them to the requests that benefit.
- Escalate on validation failure, complexity signals or explicit user choice.
- Give complete context; partial inputs waste a large model's capacity.
- Cache results for repeated premium tasks instead of regenerating.
- Track how often escalation fires — rising rates usually mean a prompt problem, not a model problem.
- Self-serve plans are flat monthly with fair-use usage — see the live pricing page for current plans.
How to switch to plugsky-max
Switching is one model name on the OpenAI-compatible endpoint: {"model": "plugsky-max", "messages": [{"role": "user", "content": "Evaluate these three architectures and recommend one with trade-offs."}], "stream": true}.
Run it in shadow mode on your hardest cases, compare against your current top model and against plugsky-frontier, then route only the qualifying traffic. Keep the fast default path intact so escalation stays the exception.
Honest comparison
| Dimension | plugsky-max | plugsky-frontier | Balanced workhorse |
|---|---|---|---|
| Best fit | Complex reasoning and long-form planning | Top-tier quality across high-value tasks | Everyday chat, agents and coding |
| Model profile | Largest general model | Frontier-profile flagship | Mid-tier workhorse |
| Latency profile | Slowest tier | Slowest tier | Balanced |
| Tools and JSON mode | Supported | Supported | Supported |
| Routing role | Escalation and deep analysis | Escalation and benchmarking | Production default |
| Failover | Automatic top-tier peer fallback | Automatic top-tier peer fallback | Automatic peer fallback |
Frequently asked questions
Is plugsky-max free?
No — it is a paid-plan model. The free plan includes two free models, plugsky-micro and plugsky-lite. A 14-day full-access trial lets you evaluate paid tiers first.
What context window does it have?
It is 128K-class today. The exact live window and output limit are published per model at /models.
Should plugsky-max be my default model?
No. It is the slowest and most expensive tier. Use it as an escalation target and keep a faster model as the default for routine traffic.
How does it differ from plugsky-frontier?
Both are top-tier general models with different profiles. Evaluate both on your hardest prompts and route per workload rather than assuming one is universally better.
Does it support tools and structured output?
Yes — streaming, function calling and JSON mode are part of the shared OpenAI-compatible surface.
When is it worth the extra latency?
When a wrong answer is costly: architecture decisions, complex analysis, difficult planning and long-form output that must remain coherent.
What happens during an upstream incident?
Requests retry through a backup upstream and same-profile top-tier peers. Live component health is on the status page.
How is pricing structured?
Self-serve plans are flat monthly with fair-use usage and no per-token billing. See the live pricing page for current plans.