Kimi Max looks slow and fragile in a three-day iOS task test
thegautam · x · 2026-07-21
Kimi gets mixed marks in a three-day iOS task test
A user says they have been running Kimi max on a highly verifiable iOS task for more than three days and reports several issues:
- signup and onboarding were easy
- the model is very slow and feels too sluggish to be load-bearing
- OpenCode follows goals better in practice
- cache/context handling is poor, with frequent context-limit problems until manual compaction is used
- guardrails appear weak, and the app is described as being one crash away from dropping f-bombs
Overall, the post reads as a negative real-world usability check rather than a benchmark claim.
More from Models
- 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
- awesome-llm-leaderboards: an open-source directory of LLM leaderboards, pricing tables, comparison tools — Last_Establishment_1 · 2026-09-11
- Anthropic claims it works to keep eval environments unidentifiable to models — MaxKannen · 2026-09-11
- Nex N2.5 Pro released on Hugging Face with 407GB of weights — jinnyjuice · 2026-09-11
- RoMa v2 image matching model unveiled in the usual black poster — ducha_aiki · 2026-09-11