Kimi feels strong, but maybe not cheap to run
JoshuaJBouw · x · 2026-07-20
The author argues that open-weight models do not make the world uniquely dangerous, because all powerful technologies are dual-use and usually end up net safer than harmful.
In a quoted section about Kimi, they add two concrete observations:
- Kimi appears to be a very strong model, and its performance does not seem easily explained away by distillation.
- In agentic coding sessions, it seems roughly on par with the best public models expected in Q1 2026.
- Their limited use suggests it is very token-hungry, and may not actually be cheap to run.
The post is therefore both a model-quality take and a practical note on its cost profile.
More from coding & agent
- GPT-6 Astra beats Factorio with enemies in 44 in-game hours at ~$4,500 API cost — liminal_bardo · 2026-09-11
- 105 hidden bugs, 2 repos: DeepSeek V4.1 Flash fixes 24 at $1.80 vs Opus 5's 27 at $51.33 — ChartsJournalX · 2026-09-11
- Investment Analyst Asks How to Build a Claude-Based Diligence Agent Stack — Careless_Tie2286 · 2026-09-11
- Treating agents like 50 First Dates: a 3-layer context system so every conversation doesn't start from zero — evielync · 2026-09-11
- Running the Firefox MCP on Android via Termux, ngrok, and mcp-proxy — Nervous-Strain7544 · 2026-09-11
- SmolVM open-sources persistent computer infrastructure for agents that outlive chat sessions — aniketmaurya · 2026-09-11