Dev Team Drops Anthropic Max for Codex and China’s Kimi, GLM, Grok in Cursor
haider1 · x · 2026-07-28
The author says their dev team used to spend hundreds of thousands of dollars on Anthropic subscriptions and API usage because there were no serious alternatives, while OpenAI was focused elsewhere.
That changed this year: they cancelled Anthropic’s Max 20x plan, moved to Codex 5x, and now use GLM-5.2, Kimi K3, and Grok 4.5 inside Cursor. The post argues Anthropic’s coding moat is shrinking fast, and the quoted reply adds that Kimi K3 just released its full weights as the world’s first open-weight model in the 2.8T-parameter class, with MiniMax expected to follow with a 2–3T open model.
Related event: Kimi K3 Challenges Anthropic's High-Cost Subscriptions(2 posts)→
More from coding & agent
- GrokTerm v0.1.11 adds two-way voice and session resume for Grok Build on macOS — Daniel_Farinax · 2026-07-28
- KAT-Coder-V2.5-Dev brings 3B active-parameter agentic coding to 256K context — airesearch12 · 2026-07-28
- Work Louder’s Codex Micro turns ChatGPT Codex into a dedicated 6-key keypad — Dimillian · 2026-07-28
- Open-source agent skill preserves the reasoning behind code changes — oliver-zehentleitner · 2026-07-28
- Anthropic’s Claude Opus 5 guide explains which prompt habits now hurt more than help — CodeByPoonam · 2026-07-28
- PIRL adds closed-loop verification to RL post-training and improves accuracy on reasoning and code tasks — This_Ad9834 · 2026-07-28