lachtan/nanobot-runtime
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

dream: 2026-06-07 14:23, 6 change(s)

Browse Source 
[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
This commit is contained in:
nanobot
2026-06-07 16:31:36 +02:00
parent 160175327c
commit 3aa1b7e990
3 changed files with 15 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
memory/MEMORY.md
View File

@@ -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