diff --git a/tiktoken_ext/openai_public.py b/tiktoken_ext/openai_public.py
index 16a6ec50..83980674 100644
--- a/tiktoken_ext/openai_public.py
+++ b/tiktoken_ext/openai_public.py
@@ -1,4 +1,5 @@
 from tiktoken.load import data_gym_to_mergeable_bpe_ranks, load_tiktoken_bpe
+import os
 
 ENDOFTEXT = "<|endoftext|>"
 FIM_PREFIX = "<|fim_prefix|>"
@@ -8,9 +9,17 @@
 
 
 def gpt2():
+    vocab_bpe_file = os.environ.get(
+        "TIKTOKEN_BPE_FILE_GPT2_VOCAB",
+        "https://openaipublic.blob.core.windows.net/gpt-2/encodings/main/vocab.bpe",
+    )
+    encoder_json_file = os.environ.get(
+        "TIKTOKEN_BPE_FILE_GPT2_ENCODER",
+        "https://openaipublic.blob.core.windows.net/gpt-2/encodings/main/encoder.json",
+    )
     mergeable_ranks = data_gym_to_mergeable_bpe_ranks(
-        vocab_bpe_file="https://openaipublic.blob.core.windows.net/gpt-2/encodings/main/vocab.bpe",
-        encoder_json_file="https://openaipublic.blob.core.windows.net/gpt-2/encodings/main/encoder.json",
+        vocab_bpe_file=vocab_bpe_file,
+        encoder_json_file=encoder_json_file,
     )
     return {
         "name": "gpt2",
@@ -22,9 +31,11 @@ def gpt2():
 
 
 def r50k_base():
-    mergeable_ranks = load_tiktoken_bpe(
-        "https://openaipublic.blob.core.windows.net/encodings/r50k_base.tiktoken"
+    tiktoken_bpe_file = os.environ.get(
+        "TIKTOKEN_BPE_FILE_R50K_BASE",
+        "https://openaipublic.blob.core.windows.net/encodings/r50k_base.tiktoken",
     )
+    mergeable_ranks = load_tiktoken_bpe(tiktoken_bpe_file)
     return {
         "name": "r50k_base",
         "explicit_n_vocab": 50257,
@@ -35,9 +46,11 @@ def r50k_base():
 
 
 def p50k_base():
-    mergeable_ranks = load_tiktoken_bpe(
-        "https://openaipublic.blob.core.windows.net/encodings/p50k_base.tiktoken"
+    tiktoken_bpe_file = os.environ.get(
+        "TIKTOKEN_BPE_FILE_P50K_BASE",
+        "https://openaipublic.blob.core.windows.net/encodings/p50k_base.tiktoken",
     )
+    mergeable_ranks = load_tiktoken_bpe(tiktoken_bpe_file)
     return {
         "name": "p50k_base",
         "explicit_n_vocab": 50281,
@@ -48,10 +61,17 @@ def p50k_base():
 
 
 def p50k_edit():
-    mergeable_ranks = load_tiktoken_bpe(
-        "https://openaipublic.blob.core.windows.net/encodings/p50k_base.tiktoken"
+    tiktoken_bpe_file = os.environ.get(
+        "TIKTOKEN_BPE_FILE_P50K_BASE",
+        "https://openaipublic.blob.core.windows.net/encodings/p50k_base.tiktoken",
     )
-    special_tokens = {ENDOFTEXT: 50256, FIM_PREFIX: 50281, FIM_MIDDLE: 50282, FIM_SUFFIX: 50283}
+    mergeable_ranks = load_tiktoken_bpe(tiktoken_bpe_file)
+    special_tokens = {
+        ENDOFTEXT: 50256,
+        FIM_PREFIX: 50281,
+        FIM_MIDDLE: 50282,
+        FIM_SUFFIX: 50283,
+    }
     return {
         "name": "p50k_edit",
         "pat_str": r"""'s|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+""",
@@ -61,9 +81,11 @@ def p50k_edit():
 
 
 def cl100k_base():
-    mergeable_ranks = load_tiktoken_bpe(
-        "https://openaipublic.blob.core.windows.net/encodings/cl100k_base.tiktoken"
+    tiktoken_bpe_file = os.environ.get(
+        "TIKTOKEN_BPE_FILE_CL100K_BASE",
+        "https://openaipublic.blob.core.windows.net/encodings/cl100k_base.tiktoken",
     )
+    mergeable_ranks = load_tiktoken_bpe(tiktoken_bpe_file)
     special_tokens = {
         ENDOFTEXT: 100257,
         FIM_PREFIX: 100258,