Fireworks AI launched day-zero US-hosted serverless inference for GLM-5.3, highlighting a 50% improvement on Code Bench and state-of-the-art results in open-weight cyber defense and long-horizon agents.

Key Takeaways

  • ✓Instant day-zero serverless API access for teams without dedicated multi-GPU infrastructure for 744B MoEs;
  • ✓Optimized specifically for complex code refactoring, cyber security benchmarks, and long-horizon tasks;
  • ✓Proprietary inference stack delivers sustained high tokens-per-second streaming performance.