title: "Subconscious is now on Vercel AI Gateway | Subconscious"

Subconscious is now available as an AI inference provider on the Vercel AI Gateway, offering models like GLM 5.3 and DeepSeek V4.1 Flash. The platform is designed for long-running agentic tasks, such as coding and cybersecurity, using runtime KV cache compression and custom speculative decoders. This technology improves speed and accuracy for sessions that consume more than 200,000 tokens. A key feature is its billing model, which only charges for tokens processed after compression, making it potentially more cost-effective for large-context requests.