Our take
Parasail is a mid-size host for downloadable models with a zero-retention data option and generally low token prices. Its 44-model catalogue suits exploratory use, though throughput varies widely and some models carry steep output-price markups.
Use this for privacy-sensitive workloads where zero-retention matters, or for low-cost inference on small-to-mid downloadable models. Pick it for exploratory access to a 44-model catalogue before committing to a larger provider. Skip it if you need verified SOC 2, a known headquarters country, or a guaranteed EU endpoint — all three are unverified in our data. Also skip it if your workload needs consistently fast throughput.
- Strong privacy option: zero-retention is available and the provider does not train on prompts.
- Very low input prices on several models — Mistral Nemo and GPT-OSS Safeguard 20B both at three cents per million tokens input.
- Broad catalogue for a non-hyperscaler: 44 tracked offers.
- Throughput is inconsistent and sometimes very low — ByteDance UI-TARS 7B at 9 TPS versus MythoMax 13B at 102 TPS.
- Several models carry steep output-price markups: Llama 3.2 3B Instruct output costs 6.6× its input price; Gemma 4 31B output costs 2.67× its input.
- Compliance and geographic documentation is thin: headquarters country, EU endpoint and SOC 2 are all unverified in our data.
Point your tools here
https://api.parasail.io/v1Model IDs on this page are LLMap's own slugs, built for browsing and comparing — not what Parasail's API expects. Its own model identifier lives in the docs above.
Privacy & data handling
These answers cover requests sent to Parasail through OpenRouter, as OpenRouter records them, last read 3 hours ago. For Parasail'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.
Prompt retention: none.