Show HN: Fast inference for deep seek flash v4.1 469 tok/s for coding

Coral Bricks, a Seattle-based inference platform founded on 2026-06-12 and backed by Afore Capital and Foundations Accelerator, launched with a claim of 469 tokens per second for DeepSeek Flash v4.1 on coding workloads. The company serves open models including Kimi, GLM and gpt-oss behind an OpenAI-compatible API, offering near-zero rate limits and free cached input tokens for long-running coding and research agents. Coral Bricks is led by CEO Hitesh Jain and Head of Engineering Divy Vasal.

Coral Bricks — High-throughput inference for heavy agents Coral Bricks is an inference platform for coding and research agents that plan, call tools and reason over large contexts. It serves open models such as Kimi, GLM and gpt-oss behind an OpenAI-compatible API, with multiple times the tokens per second of a typical provider, near-zero rate limits, and cached input tokens free — so long-running agent workloads finish on time instead of queuing. Company - Founded: 2026-06-12 - Headquarters: Seattle, WA, US - Backed by: Afore Capital and Foundations Accelerator - Contact: hello@coralbricks.ai mailto:hello@coralbricks.ai Founders - Hitesh Jain — CEO - Divy Vasal — Head of Engineering Links