Our take
Upstage is a small, focused inference host built around a single model family. Its latest Solar Pro 4 generation is sharply cheaper than its predecessor, with explicit data-privacy controls including a zero-retention option.
Use this for budget-conscious inference on the Solar Pro 4 generation, or when you need a zero-retention data option. Skip it if you need a broad model catalogue, verified SOC 2, a confirmed EU endpoint, or consistent throughput guarantees.
- Latest-generation Solar Pro 4 is sharply cheaper than its predecessor from the same provider — a 5× price reduction on both input and output.
- Does not train on prompts and offers a zero-retention option.
- Pro 4 shows moderate throughput at 33–35 tokens per second on two endpoints; Pro 3 reaches 120 tokens per second on one endpoint.
- Very narrow model catalogue — only 5 tracked offers, all from a single model family.
- Compliance and geographic footprint are undocumented: SOC 2, headquarters country and EU-region endpoint are all unverified in our data.
- Throughput is inconsistent and sometimes very low: Pro 4 drops to 10 tokens per second on one endpoint, and Pro 3 falls to 4 tokens per second on another.
Point your tools here
We have not recorded what a router needs for Upstage yet — no base URL and no docs link on file. That is our gap, not a sign Upstage has no API; its own documentation is the place to look until we close it.
Privacy & data handling
These answers cover requests sent to Upstage through OpenRouter, as OpenRouter records them, last read 58 min ago. For Upstage's own API we hold no answer: check its terms. SOC 2 and the data processing agreement below are hand-checked, not part of that sync.