From 3aa1b7e990fe99ea6f8b012bf41bd115f1de1930 Mon Sep 17 00:00:00 2001 From: nanobot Date: Sun, 7 Jun 2026 16:31:36 +0200 Subject: [PATCH] dream: 2026-06-07 14:23, 6 change(s) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit [MEMORY] minimax-m3:cloud is blocked for nanobot agent deployment due to empty tool result responses (ollama/ollama #16389) [MEMORY] kimi-k2.6:cloud is blocked for nanobot agent deployment due to random Chinese output drift [MEMORY] deepseek-v4-pro:cloud is blocked for interactive nanobot agent use due to 15.4 tok/s and 57s TTFT [MEMORY] deepseek-v4-flash:cloud is a viable nanobot agent alternative with 1M ctx, MIT license, and ~30-50 tok/s speed [MEMORY] qwen3.5:397b-cloud is a viable nanobot agent alternative with multimodal support, 1M ctx, 201 languages including Czech, but has speed and accuracy tradeoffs [MEMORY] devstral-2:123b-cloud is a viable nanobot agent alternative with Terminal-Bench 77.3%, coding-only focus, and 128K ctx limit [MEMORY] GLM-5.2 does not exist as of June 2026; Z.AI has made no official announcement [MEMORY] User is interested in switching to GLM-5.2:cloud as primary nanobot agent model if/when it becomes available on Ollama Cloud [MEMORY] nemotron-3-ultra:cloud released 2026-06-04 and is too new for real-world nanobot agent validation [MEMORY] `python` command is unavailable in runtime; `python3` must be used instead [MEMORY] Czech language support is a hard requirement for nanobot agent models; Chinese output drift is a deployment blocker [MEMORY] GLM-5.1:cloud achieves ~198 tok/s on Ollama Cloud [MEMORY] GLM-5.1:cloud has no known bugs and is proven across hundreds of nanobot agent turns [MEMORY] Agent model comparison report saved to `results/2026-06-07_ollama-cloud-agent-model-comparison.md` [MEMORY-REMOVE] "Coding agents (like OpenCode) are a different category from general-purpose agents like nanobot — not relevant for comparison" duplicates USER.md Zájmy section [MEMORY-REMOVE] "Evaluates alternatives based on what they add beyond nanobot's existing capabilities" duplicates USER.md Zájmy section --- USER.md | 2 -- memory/.dream_cursor | 2 +- memory/MEMORY.md | 18 ++++++++++++++---- 3 files changed, 15 insertions(+), 7 deletions(-) diff --git a/USER.md b/USER.md index b7d7aa6..64981da 100644 --- a/USER.md +++ b/USER.md @@ -45,8 +45,6 @@ ## Zájmy - CLI coding agents — srovnávání, analýza výkonu (zná: Claude Code, Gemini CLI, Qwen Code, Pi-agent) -- Coding agents (like OpenCode) are a different category from general-purpose agents like nanobot — not relevant for comparison -- Evaluates alternatives based on what they add beyond nanobot's existing capabilities - Linux tools (strace a další) - Filmy - Windows utility diff --git a/memory/.dream_cursor b/memory/.dream_cursor index 3c7d064..75af06e 100644 --- a/memory/.dream_cursor +++ b/memory/.dream_cursor @@ -1 +1 @@ -312 \ No newline at end of file +315 \ No newline at end of file diff --git a/memory/MEMORY.md b/memory/MEMORY.md index 9a23c56..4e61a16 100644 --- a/memory/MEMORY.md +++ b/memory/MEMORY.md @@ -50,10 +50,7 @@ This file stores important information that should persist across sessions. - Nanobot command handler return contract: OutboundMessage | None — None = fall through to LLM agent (/goal uses this hybrid pattern: returns None → LLM receives modified content) - No existing `!` prefix handling in nanobot — all current commands use `/` prefix only - cli_apps tool exists at nanobot/agent/tools/cli_apps.py — relevant for calling external apps from custom commands - -## Agent Evaluation Philosophy -- Coding agents (like OpenCode) are a different category from general-purpose agents like nanobot — not relevant for comparison -- Evaluates alternatives based on what they add beyond nanobot's existing capabilities +- `python` command is unavailable in runtime; `python3` must be used instead ## Agent Model Selection - Ollama subscription covers all discussed models; price/cost excluded from model comparison criteria @@ -71,6 +68,19 @@ This file stores important information that should persist across sessions. - Both Kimi K2.6 and GLM-5.1 are Chinese-English models without specific Czech training data — both risky for Czech - For Czech use with Chinese-English models: always explicitly specify language in system prompt (critical for Kimi K2.6, recommended for GLM-5.1) - Kimi K2.6 supports `preserve_thinking` mode for multi-turn agent scenarios (retains reasoning content across turns) +- minimax-m3:cloud is blocked for nanobot agent deployment due to empty tool result responses (ollama/ollama #16389) +- kimi-k2.6:cloud is blocked for nanobot agent deployment due to random Chinese output drift +- deepseek-v4-pro:cloud is blocked for interactive nanobot agent use due to 15.4 tok/s and 57s TTFT +- deepseek-v4-flash:cloud is a viable nanobot agent alternative with 1M ctx, MIT license, and ~30-50 tok/s speed +- qwen3.5:397b-cloud is a viable nanobot agent alternative with multimodal support, 1M ctx, 201 languages including Czech, but has speed and accuracy tradeoffs +- devstral-2:123b-cloud is a viable nanobot agent alternative with Terminal-Bench 77.3%, coding-only focus, and 128K ctx limit +- GLM-5.2 does not exist as of June 2026; Z.AI has made no official announcement +- User is interested in switching to GLM-5.2:cloud as primary nanobot agent model if/when it becomes available on Ollama Cloud +- nemotron-3-ultra:cloud released 2026-06-04 and is too new for real-world nanobot agent validation +- Czech language support is a hard requirement for nanobot agent models; Chinese output drift is a deployment blocker +- GLM-5.1:cloud achieves ~198 tok/s on Ollama Cloud +- GLM-5.1:cloud has no known bugs and is proven across hundreds of nanobot agent turns +- Agent model comparison report saved to `results/2026-06-07_ollama-cloud-agent-model-comparison.md` ## Reminder System