INT21's AI agent swarm rewrote Drama 3's inference runtime in Rust within one hour, beating SGLang-Omni
bingxu_ · x · 2026-10-26
Infrastructure startup INT21 says its self-improving AI agent swarm built a custom Rust inference runtime for Fish Audio's Drama 3 model in about an hour — without early access to the model — and outperformed the SGLang-Omni framework.
Key points from the wccftech interview with CEO Bing Xu:
- SwarmOS orchestrates thousands of AI agents inside an "Inference Engine Factory" that writes, tests, and optimizes runtime code autonomously
- The stack drops Python entirely in favor of Rust to eliminate interpreter overhead
- Core thesis: "buying more B200s and fixing runtimes are orthogonal" — compute should be used to improve compute, and software optimization matters more than more GPUs
- Xu claims the agent swarm out-coded human experts
More from coding & agent
- Firecrawl offers 2.5x credits to a16z Speedrun founders, unveils $75M Series B — devdigest · 2026-10-27
- Chums: open-source Apple Watch app puts Claude Code and Codex on your wrist — brandon_galang · 2026-10-27
- Open-source Clairic Studio turns pasted text into typed data and working code in 5 languages — dfinke · 2026-10-27
- Matt Pocock shares expand/contract trick: split one-way-door PRs to de-risk big changes — mattpocockuk · 2026-10-27
- Qdrant builds Universal I/O to run vector search mmap code on object storage — qdrant_engine · 2026-10-27
- theo publishes tstack skills, Garry Tan urges everyone to 'publish your stack' — ycombinator · 2026-10-26