Alchemy launches inference API offering open-weight models with low latency, low costs, and zero data retention

Alchemy has launched an inference API offering open-weight models including Kimi, DeepSeek, and GLM through an OpenAI-compatible endpoint. The company says the service provides lower costs without compromising performance, low latency with high throughput, reliability backed by Alchemy's 99.99% uptime record, and zero data retention. According to the announcement, developers can switch by pointing their existing SDK at the new endpoint, swapping their key, and picking a model, with no new SDKs or GPU management required. The company also states privacy should not be an upsell.

✨ AI summary of the linked source — not editorial content.

Explore the topic

• Launch
• API
• Alchemy