Our take
AkashML is a small inference host with a zero-retention policy and aggressive pricing on a handful of models. It offers a 10-model catalogue with throughput-versus-price tiers on several downloads.
Use this when zero data retention is required and your model is among the 10 tracked offers. Pick it for high-throughput runs on specific Qwen or Z-AI GLM variants where 93–112.5 tps is available, or for the cheapest tracked GPT-OSS 120B input price. Skip it if you need verified SOC 2, a confirmed EU endpoint, or a broad model catalogue.
- Zero-retention option available, with a commitment not to train on prompts.
- Cheapest tracked input price for GPT-OSS 120B in this catalogue.
- Highest throughput option for Z-AI GLM 5-3 at 112.5 tps.
- Price-performance tiers on some models: DeepSeek V4 Flash input is 2.4× cheaper at the 17 tps tier than the 7 tps tier.
- Compliance and geographic posture largely undocumented: headquarters, EU endpoint and SOC 2 are all unverified in our data.
- Small catalogue — only 10 tracked offers.
- Inconsistent throughput-to-price relationship: on DeepSeek V4 Flash you pay 51% more input cost for 59% lower throughput, while on Qwen3-8 27B you pay 14% more for 16% lower throughput.
Point your tools here
We have not recorded what a router needs for AkashML yet — no base URL and no docs link on file. That is our gap, not a sign AkashML has no API; its own documentation is the place to look until we close it.
Privacy & data handling
These answers cover requests sent to AkashML through OpenRouter, as OpenRouter records them, last read 58 min ago. For AkashML'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.