GLM 5.3 hits 200+ TPS on Together API, codes a new webpage in 37 seconds
nutlope · x · 2026-09-03
Developer Nutlope demoed GLM 5.3 running at over 200 tokens per second on the Together API (also visible on OpenRouter). An unsped-up screen recording shows the model adding a complete new page to his personal site in just 37 seconds, highlighting its coding speed.
Related event: GLM 5.3 hits 200+ TPS, builds full webpage in 37 seconds(2 posts)→
More from coding & agent
- HarnessEvolve paper: dual-gate loop fixes three failure modes of self-evolving agents — dair_ai · 2026-09-03
- Mastra launches sandbox computer use so agents can browse, click, type and screenshot — Scobleizer · 2026-09-03
- How to Keep Long-Running Agents on Track: Redis State Beats Context Stuffing — Deepfeet-09 · 2026-09-03
- Looking for a <500M SLM to Summarize Code Snippets for Local Agents — Mrinohk · 2026-09-03
- LangSmith Roadshow heads to Dallas in two weeks with agent dev workshops — LangChain · 2026-09-03
- Agentic API adds a stateful layer in front of vLLM for open-model agent runtimes — techNmak · 2026-09-03