FlashLabs enhances OrcaRouter with post-quantum cryptography and new AI models

FlashLabs has announced several major updates to OrcaRouter, an AI inference gateway developed by Continuum AI and distributed exclusively by FlashLabs in Japan.

To combat the “harvest now, decrypt later” threat, OrcaRouter now features standard, no-cost post-quantum cryptography for all request logs. This system uses a hybrid configuration of ML-KEM-768 and X25519 to protect data before it reaches storage, ensuring that even if databases or backups are compromised, the contents remain unreadable. The underlying implementation, “scuttle,” has been released as open-source under the Apache-2.0 license.

Additionally, FlashLabs introduced “Jev 1.13,” a specialized AI model designed for rapid, structured decision-making rather than text generation. It provides responses in JSON format for tasks like classification and verification, significantly reducing latency and costs for AI agents. The company also released “OrcaSAQ-2 27B,” a highly compressed version of the Qwen3.8-27B model. This model reduces file size from 54GB to 12.3GB while maintaining a 93.2% response consistency rate with the original, making it more accessible for local GPU environments.

Finally, FlashLabs will participate in the “TiDB User Day 2026” in Tokyo, where they will present research on model access design and automated routing for the agentic AI era.