The last task was from 23:29/11:29 PM to 23:59/11:59 PM with GPT 5.6 Sol xHigh
Total= 10min + 30min = 40 min
Start: 11:17 - 74%
Finish: 11:59 - 73%
Result: -1% in 40min de with GPT 5.6 Sol xHigh in Codex Local
statistically significant. Context without wrapper: 89 thousand (medium) and 155 thousand (large). With wrapper: 22 thousand and 32 thousand. Strong gain on medium and large—not every case. Learn more: https://t.co/IKKM0RQkyQ
Real case, 17 Sep 2026: Pasblinn running Simplicio Runtime 3.8.53 on OpenCode with qwen/qwen3.7-flash on Alibaba. Real cost on OpenRouter: medium prompt $0.11890 → $0.00999 (−91.6%); large prompt $0.20947 → $0.04147 (−80.2%); small prompt $0.00462 → $0.00522 (+13.2%), not
Aprendizados da noite na aula do maior contribuidor BR do Hermes @simpliciomcp@wesleysimplic
- Agentes criam arquivos temporários locais chamados de artefatos para se guiarem (normalmente em SQLlite)
- A mina de ouro está no cache hit
- Conceito e funcionamento de RAG
- RAG x Prompt
- tipos de orquestração: GOAL x LOOP x WORKFLOW
New release
3.8.32
8 threads > 20 > 40
32 agents
Bug fixes - pre hooks Claude and Codex
Depois fale pra ele rodar:
macOS / Linux:
python3 -m pip install --upgrade simplicio-installer
simplicio install
The last task was from 23:29/11:29 PM to 23:59/11:59 PM with GPT 5.6 Sol xHigh
Total= 10min + 30min = 40 min
Start: 11:17 - 74%
Finish: 11:59 - 73%
Result: -1% in 40min de with GPT 5.6 Sol xHigh in Codex Local
The last task was from 23:29/11:29 PM to 23:59/11:59 PM with GPT 5.6 Sol xHigh
Total= 10min + 30min = 40 min
Start: 11:17 - 74%
Finish: 11:59 - 73%
Result: -1% in 40min de with GPT 5.6 Sol xHigh in Codex Local