plugsky-frontier — frontier quality, sub-second

Summary

Model	Identity	Context	Latency*	Best for
plugsky-frontier	Mistral Large 3 (675B)	128K tokens	~0.70s median	Frontier reasoning, enterprise workloads, the hardest tasks

*Median measured latency on the live API at https://api.plugsky.com/v1.

Specs

Property	Value
Model family	Mistral Large 3 (675B, EU-origin)
Context window	128K tokens
Measured latency	~0.70s median on the live API
Capabilities	Frontier reasoning, tool calling, long context
Available on	Scale plan and above

When to use plugsky-frontier

• Frontier tasks — the 5% of prompts where quality decides the product.
• Enterprise workloads — EU-origin model for EU data-flows.
• Reasoning under pressure — multi-step, high-stakes output.

FAQ

Is it really a 675B model?

Yes — Mistral Large 3 675B, hosted with EU-origin infrastructure.

How is latency so low at 675B?

Fast inference infrastructure and MoE efficiency; measured median is ~0.70s on the public API.

Can enterprises deploy it in-region?

Yes — see enterprise deployment for private endpoints and on-prem options.