diff --git a/README.md b/README.md
index 1e4467e74..b45719975 100644
--- a/README.md
+++ b/README.md
@@ -361,7 +361,7 @@ These are only needed for **headless / CI extraction** (`graphify extract`). Whe
 | `AWS_*` / `~/.aws/credentials` | AWS Bedrock — standard credential chain | `--backend bedrock` (no API key, uses IAM) |
 | `GRAPHIFY_MAX_WORKERS` | AST parallelism thread count | optional — also `--max-workers` flag |
 | `GRAPHIFY_MAX_OUTPUT_TOKENS` | Raise output cap for dense corpora | optional — e.g. `32768` for large files |
-| `GRAPHIFY_API_TIMEOUT` | HTTP timeout in seconds (default: 600) | optional — also `--api-timeout` flag |
+| `GRAPHIFY_API_TIMEOUT` | Per-call timeout in seconds for every backend, HTTP and `claude-cli` alike (default: 600) | optional — also `--api-timeout` flag |
 | `GRAPHIFY_FORCE` | Force graph rebuild even with fewer nodes | optional — also `--force` flag |
 | `GRAPHIFY_GOOGLE_WORKSPACE` | Auto-enable Google Workspace export | optional — set to `1` |
 | `GRAPHIFY_TRIAGE_BACKEND` | Backend for `graphify prs --triage` | optional — auto-detected from available keys |
diff --git a/graphify/llm.py b/graphify/llm.py
index c67014b4a..407fda538 100644
--- a/graphify/llm.py
+++ b/graphify/llm.py
@@ -157,6 +157,25 @@ def _resolve_max_tokens(default: int) -> int:
             pass
     return default
 
+
+def _resolve_api_timeout(default: float = 600.0) -> float:
+    """Honour GRAPHIFY_API_TIMEOUT env var override (seconds), else use the default.
+
+    Applies uniformly across every backend — the OpenAI-compatible HTTP client
+    and the claude-cli subprocess paths alike — so a slow local model or a long
+    chunk can be given more headroom without patching the source. Non-positive
+    or unparseable values fall back to the default (issue #792 addendum).
+    """
+    raw = os.environ.get("GRAPHIFY_API_TIMEOUT", "").strip()
+    if raw:
+        try:
+            v = float(raw)
+            if v > 0:
+                return v
+        except ValueError:
+            pass
+    return default
+
 _EXTRACTION_SYSTEM = """\
 You are a graphify semantic extraction agent. Extract a knowledge graph fragment from the files provided.
 Output ONLY valid JSON — no explanation, no markdown fences, no preamble.
@@ -362,19 +381,10 @@ def _call_openai_compat(
 
     # Local backends (ollama, llama.cpp, vLLM) routinely take >60s for a
     # single chunk on a large model — far longer than the openai SDK's
-    # default. Honour GRAPHIFY_API_TIMEOUT (seconds) for explicit override;
-    # default to 600s, which is long enough for a 31B model on a 16k chunk
-    # but still bounds runaway connections (issue #792 addendum).
-    timeout_raw = os.environ.get("GRAPHIFY_API_TIMEOUT", "").strip()
-    timeout_s: float = 600.0
-    if timeout_raw:
-        try:
-            v = float(timeout_raw)
-            if v > 0:
-                timeout_s = v
-        except ValueError:
-            pass
-    client = OpenAI(api_key=api_key, base_url=base_url, timeout=timeout_s)
+    # default. GRAPHIFY_API_TIMEOUT (seconds) overrides; defaults to 600s,
+    # long enough for a 31B model on a 16k chunk but still bounding a runaway
+    # connection (issue #792 addendum).
+    client = OpenAI(api_key=api_key, base_url=base_url, timeout=_resolve_api_timeout())
     kwargs: dict = {
         "model": model,
         "messages": [
@@ -571,7 +581,7 @@ def _call_claude_cli(user_message: str, max_tokens: int = 8192, *, deep_mode: bo
         capture_output=True,
         text=True,
         encoding="utf-8",  # Force UTF-8 — prevents UnicodeEncodeError on Windows cp1252
-        timeout=600,
+        timeout=_resolve_api_timeout(),
         check=False,
     )
     if proc.returncode != 0:
@@ -1134,7 +1144,7 @@ def _call_llm(prompt: str, *, backend: str, max_tokens: int = 200) -> str:
             capture_output=True,
             text=True,
             encoding="utf-8",  # Force UTF-8 — prevents UnicodeEncodeError on Windows cp1252
-            timeout=600,
+            timeout=_resolve_api_timeout(),
             check=False,
         )
         if proc.returncode != 0:
diff --git a/tests/test_claude_cli_backend.py b/tests/test_claude_cli_backend.py
index eeb6fd27b..f204ee53a 100644
--- a/tests/test_claude_cli_backend.py
+++ b/tests/test_claude_cli_backend.py
@@ -193,3 +193,38 @@ def test_non_windows_uses_bare_claude(monkeypatch):
 
     argv = run.call_args.args[0]
     assert argv[0] == "claude"
+
+
+# ---------- timeout honouring: GRAPHIFY_API_TIMEOUT reaches claude-cli ----------
+
+
+def test_resolve_api_timeout_default(monkeypatch):
+    monkeypatch.delenv("GRAPHIFY_API_TIMEOUT", raising=False)
+    assert llm._resolve_api_timeout() == 600.0
+
+
+def test_resolve_api_timeout_env_override(monkeypatch):
+    monkeypatch.setenv("GRAPHIFY_API_TIMEOUT", "45")
+    assert llm._resolve_api_timeout() == 45.0
+
+
+def test_resolve_api_timeout_ignores_invalid(monkeypatch):
+    monkeypatch.setenv("GRAPHIFY_API_TIMEOUT", "not-a-number")
+    assert llm._resolve_api_timeout() == 600.0
+
+
+def test_resolve_api_timeout_ignores_nonpositive(monkeypatch):
+    monkeypatch.setenv("GRAPHIFY_API_TIMEOUT", "0")
+    assert llm._resolve_api_timeout() == 600.0
+
+
+def test_claude_cli_subprocess_uses_default_timeout(fake_claude, monkeypatch):
+    monkeypatch.delenv("GRAPHIFY_API_TIMEOUT", raising=False)
+    llm._call_claude_cli("dummy", max_tokens=8192)
+    assert fake_claude.call_args.kwargs["timeout"] == 600.0
+
+
+def test_claude_cli_subprocess_honours_api_timeout_env(fake_claude, monkeypatch):
+    monkeypatch.setenv("GRAPHIFY_API_TIMEOUT", "30")
+    llm._call_claude_cli("dummy", max_tokens=8192)
+    assert fake_claude.call_args.kwargs["timeout"] == 30.0