Recognise gpt-5.1 model identifier (Fixes #464)

tests/test_misc.py

@@ -19,6 +19,12 @@ def test_encoding_for_model():
     assert enc.name == "o200k_base"
     enc = tiktoken.encoding_for_model("gpt-oss-120b")
     assert enc.name == "o200k_harmony"
+    # Regression for #464: a dotted minor like 'gpt-5.1' does not start
+    # with the 'gpt-5-' prefix, so it needs its own entry / prefix.
+    enc = tiktoken.encoding_for_model("gpt-5.1")
+    assert enc.name == "o200k_base"
+    enc = tiktoken.encoding_for_model("gpt-5.1-2025-11")
+    assert enc.name == "o200k_base"
 
 
 def test_optional_blobfile_dependency():

tiktoken/model.py

@@ -9,6 +9,12 @@
     "o3-": "o200k_base",
     "o4-mini-": "o200k_base",
     # chat
+    # NB: list 'gpt-5.1-' before 'gpt-5-' so that a dotted minor like
+    # 'gpt-5.1-2025-11' is matched by its own prefix rather than failing the
+    # 'gpt-5-' check ('gpt-5.1-' does not start with 'gpt-5-'). For dict
+    # iteration order this is purely cosmetic — startswith() is correct either
+    # way — but it makes the intent explicit.
+    "gpt-5.1-": "o200k_base",
     "gpt-5-": "o200k_base",
     "gpt-4.5-": "o200k_base",
     "gpt-4.1-": "o200k_base",
@@ -32,6 +38,7 @@
     "o3": "o200k_base",
     "o4-mini": "o200k_base",
     # chat
+    "gpt-5.1": "o200k_base",
     "gpt-5": "o200k_base",
     "gpt-4.1": "o200k_base",
     "gpt-4o": "o200k_base",