~95% on SimpleQA (e.g. Qwen3.6-27B on a 3090). Supports all local and cloud LLMs (llama.cpp, Ollama, Google, ...). 10+ search engines - arXiv, PubMed, your private documents. Everything Local & Encrypted.
Python academiaanthropicarxivbravedeep-researchencryptionhome-automationhomeserverlocallocal-deep-researchlocal-llmmistralollamaopenaipubmedresearchresearch-toolretrieval-augmented-generationsearxngself-hosted
chore(settings): bump default local context window 18432 -> 30208 (#3669)
LearningCircuit15ae51be9f66
diff --git a/src/local_deep_research/defaults/default_settings.json b/src/local_deep_research/defaults/default_settings.jsonindex 6fc5f7a25c..1754974b19 100644--- a/src/local_deep_research/defaults/default_settings.json+++ b/src/local_deep_research/defaults/default_settings.json@@ -452,7 +452,7 @@},"llm.local_context_window_size": {"category": "llm_parameters",- "description": "Context window size in tokens for local LLMs (Ollama, LlamaCpp, LM Studio). The default of 18432 tokens is recommended for the langgraph-agent strategy which needs space for tool call history. If you have limited VRAM, reduce this and switch to the source-based strategy instead.",+ "description": "Context window size in tokens for local LLMs (Ollama, LlamaCpp, LM Studio). The default of 30208 tokens leaves more headroom for the synthesis prompt (sources + previous knowledge + instructions) so it is less likely to be truncated by the model. If you have limited VRAM, reduce this and switch to the source-based strategy instead.","editable": true,"max_value": 131072.0,"min_value": 512.0,@@ -461,7 +461,7 @@"step": 512.0,"type": "LLM","ui_element": "number",- "value": 18432,+ "value": 30208,"visible": true},"llm.supports_max_tokens": {diff --git a/tests/settings/golden_master_settings.json b/tests/settings/golden_master_settings.jsonindex df44acedcc..aa3dc9703b 100644--- a/tests/settings/golden_master_settings.json+++ b/tests/settings/golden_master_settings.json@@ -1007,7 +1007,7 @@},"llm.local_context_window_size": {"category": "llm_parameters",- "description": "Context window size in tokens for local LLMs (Ollama, LlamaCpp, LM Studio). The default of 18432 tokens is recommended for the langgraph-agent strategy which needs space for tool call history. If you have limited VRAM, reduce this and switch to the source-based strategy instead.",+ "description": "Context window size in tokens for local LLMs (Ollama, LlamaCpp, LM Studio). The default of 30208 tokens leaves more headroom for the synthesis prompt (sources + previous knowledge + instructions) so it is less likely to be truncated by the model. If you have limited VRAM, reduce this and switch to the source-based strategy instead.","editable": true,"max_value": 131072.0,"min_value": 512.0,@@ -1016,7 +1016,7 @@"step": 512.0,"type": "LLM","ui_element": "number",- "value": 18432,+ "value": 30208,"visible": true},"llm.max_tokens": {