diff --git a/README.md b/README.md
index f5d9ef748..e4c4b0e7a 100644
--- a/README.md
+++ b/README.md
@@ -226,6 +226,33 @@ The MCP server gives your assistant structured access: `query_graph`, `get_node`
 
 ---
 
+## Using Kimi or a custom "cheaper" LLM instead of Claude
+
+Graphify can use **Kimi 2.6** (Via Moonshot) or any other openai-compatible model for semantic extraction. Kimi, for example is roughly 3x cheaper than claude per token.
+
+### Option A: Kimi 2.6 via Moonshot (recommended)
+
+Get a key from https://platform.moonshot.ai/, configure it as an env var, No further config needed.
+
+```bash
+export MOONSHOT_API_KEY=<your-moonshot-key>
+```
+
+### Option B: Any other LLM provider
+
+Graphify can use any LLM provider which gives an openAI compatible endpoint:
+
+Example with Openrouter and Qwen3.6
+```bash
+export GRAPHIFY_LLM_BASE_URL=https://openrouter.ai/api/v1
+export GRAPHIFY_LLM_MODEL=qwen/qwen3.6-35b-a3b
+export GRAPHIFY_LLM_API_KEY=<your-openrouter-key>
+# Optional, if required by your model
+export GRAPHIFY_LLM_TEMPERATURE=0.6
+```
+
+---
+
 ## Full command reference
 
 ```
diff --git a/graphify/llm.py b/graphify/llm.py
index 9cba0ff81..282353607 100644
--- a/graphify/llm.py
+++ b/graphify/llm.py
@@ -58,6 +58,14 @@ def _get_tokenizer():
         "pricing": {"input": 0.74, "output": 4.66},  # USD per 1M tokens
         "temperature": None,  # kimi-k2.6 enforces its own fixed temperature; sending any value raises 400
     },
+    "custom": {
+        # For users who want to specify their own OpenAI-compatible endpoint and model.
+        "base_url": os.environ.get("GRAPHIFY_LLM_BASE_URL", None),  # Must be set by caller
+        "default_model": os.environ.get("GRAPHIFY_LLM_MODEL", "kimi-k2.5"),  # Must be set by caller
+        "env_key": "GRAPHIFY_LLM_API_KEY",
+        "pricing": {"input": 0.0, "output": 0.0},  # User must estimate separately
+        "temperature": os.environ.get("GRAPHIFY_LLM_TEMPERATURE", None),  # Optional override; defaults to None (use model default)
+    }
 }
 
 _EXTRACTION_SYSTEM = """\
@@ -468,11 +476,14 @@ def estimate_cost(backend: str, input_tokens: int, output_tokens: int) -> float:
 def detect_backend() -> str | None:
     """Return the name of whichever backend has an API key set, or None.
 
-    Kimi is checked first (opt-in). Falls back to Claude if ANTHROPIC_API_KEY is set.
-    Claude is the default for the skill.md subagent pipeline and is never forced here.
+    Kimi is checked first (opt-in). Then checks for a custom defined LLM with Base URL. 
+    Falls back to Claude if ANTHROPIC_API_KEY is set. Claude is the default for the skill.md 
+    subagent pipeline and is never forced here.
     """
     if os.environ.get("MOONSHOT_API_KEY"):
         return "kimi"
+    if os.environ.get("GRAPHIFY_LLM_BASE_URL"):
+        return "custom"
     if os.environ.get("ANTHROPIC_API_KEY"):
         return "claude"
     return None