Products & ServicesEvent: September 1, 2026
Makora launched GLM-5.3-Flash, a 320B MoE, 18B active per token, 1M context model, available via serverless or as a dedicated endpoint.
Published
Signal category
Products & Services
Quote
“GLM-5.3-Flash is now live on Makora, alongside GLM-5.3.”
— Makora team
Company
Makora
AI-powered GPU kernel generation and optimization
- Industry
- Software Development
- Location
- New York, US
- Company size
- 44 employees
Our technology eliminates the need for expensive and manual GPU optimization. MakoraGenerate is an AI agent that writes and validates GPU kernels in CUDA, Triton, and more. MakoraOptimize automatically tunes hyperparameters in vLLM and SGlang to maximize performance. Makora was formerly known as Mako.
Founded 2024