Our take
StreamLake is a mid-size host for downloadable models with 27 tracked offers, competitive pricing on DeepSeek and Qwen models, and standout throughput on one coding specialist. Its compliance documentation is largely unverified.
Use this for budget-conscious inference on DeepSeek V4 Flash when the cheaper of its two listings is available. Pick it for high-throughput coding workloads on Qwen3 Coder Next, or for exploratory access to niche Chinese-ecosystem models not widely hosted elsewhere. Skip it if your workload requires verified SOC 2, a known headquarters country, or a guaranteed EU data-residency endpoint.
- Lowest-priced tracked offer for DeepSeek V4 Flash among its own listings — an 8.3% cheaper input rate than its alternate listing.
- Exceptional throughput on Qwen3 Coder Next at 100.5 tokens per second, roughly double the next-best in its own catalogue.
- Broad model diversity including hard-to-find Chinese lab models: Xiaomi MiMo, Z-AI GLM, and Kwaipilot Kat Coder among 27 offers.
- Compliance and data-residency documentation largely absent: headquarters country, EU endpoint, SOC 2, and zero-retention all unverified in our data.
- Highly variable throughput with several slow listings: Qwen3 30B A3B at 14 tokens per second is 3.4× slower than its own Qwen3 Coder Next.
- Noisy duplicate listings and steep output-token premiums: two DeepSeek V4 Flash entries with a 10.4% spread, and DeepSeek V3 at a 4.0× output-to-input ratio.
Point your tools here
We have not recorded what a router needs for StreamLake yet — no base URL and no docs link on file. That is our gap, not a sign StreamLake has no API; its own documentation is the place to look until we close it.
Privacy & data handling
These answers cover requests sent to StreamLake through OpenRouter, as OpenRouter records them, last read 4 hours ago. For StreamLake'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.