Kimi K3 took 75 minutes and still failed a simple diagram task, user says
MinusKarma01 · reddit · 2026-07-21
The poster tried to use Kimi K3 through kimi-cli and the official API for an architecture-diagram task, with only $2 of credit left.
- Kimi K3 spent about 30 minutes thinking and then another 45 minutes building, with a network error along the way.
- The run ended at negative balance (-$0.92) and still did not finish.
- The same task on DeepSeek V4 Pro via opencode produced an acceptable result in a few minutes, cost $0.16, and still allowed a later refactor.
- The poster says the hand-finished remainder was still faster than waiting for Kimi K3, and asks whether anyone is actually using it successfully.
More from coding & agent
- Hugging Face ships hf CLI updates: deduped cache saves 19GB, faster buckets, retryable GPU jobs — yb2698 · 2026-10-26
- Fern 0.1 debuts: a model built to audit training data and debug agent traces for AI builders — bneyshabur · 2026-10-26
- Fern 0.1 posts 73.4 on SWE-Bench Verified, far behind frontier models on agentic benchmarks — bneyshabur · 2026-10-26
- One prompt, full Blender animation: dev says he never touched Blender once — iamfakhrealam · 2026-10-26
- Claude Code finds root backdoor in Honda Civic head unit, then debloats it — IndraVahan · 2026-10-26
- Claude Code + Monid MCP builds a $12K/year B2B prospecting engine on live data — aryanXmahajan · 2026-10-26