unique_toolkit 0.8.8__py3-none-any.whl → 0.8.9__py3-none-any.whl

unique_toolkit/language_model/infos.py

@@ -507,7 +507,9 @@ class LanguageModelInfo(BaseModel):
                     version="claude-3-7-sonnet",
                     encoder_name=EncoderName.O200K_BASE,  # TODO: Update encoder with litellm
                     token_limits=LanguageModelTokenLimits(
-                        token_limit_input=200_000, token_limit_output=128_000
+                        # Input limit is 200_000, we leave 20_000 tokens as buffer due to tokenizer mismatch
+                        token_limit_input=180_000,
+                        token_limit_output=128_000,
                     ),
                     info_cutoff_at=date(2024, 10, 31),
                     published_at=date(2025, 2, 24),
@@ -525,7 +527,9 @@ class LanguageModelInfo(BaseModel):
                     version="claude-3-7-sonnet-thinking",
                     encoder_name=EncoderName.O200K_BASE,  # TODO: Update encoder with litellm
                     token_limits=LanguageModelTokenLimits(
-                        token_limit_input=200_000, token_limit_output=128_000
+                        # Input limit is 200_000, we leave 20_000 tokens as buffer due to tokenizer mismatch
+                        token_limit_input=180_000,
+                        token_limit_output=128_000,
                     ),
                     info_cutoff_at=date(2024, 10, 31),
                     published_at=date(2025, 2, 24),
@@ -543,7 +547,9 @@ class LanguageModelInfo(BaseModel):
                     version="claude-sonnet-4",
                     encoder_name=EncoderName.O200K_BASE,  # TODO: Update encoder with litellm
                     token_limits=LanguageModelTokenLimits(
-                        token_limit_input=200_000, token_limit_output=64_000
+                        # Input limit is 200_000, we leave 20_000 tokens as buffer due to tokenizer mismatch
+                        token_limit_input=180_000,
+                        token_limit_output=64_000,
                     ),
                     info_cutoff_at=date(2025, 3, 1),
                     published_at=date(2025, 5, 1),
@@ -561,7 +567,9 @@ class LanguageModelInfo(BaseModel):
                     version="claude-opus-4",
                     encoder_name=EncoderName.O200K_BASE,  # TODO: Update encoder with litellm
                     token_limits=LanguageModelTokenLimits(
-                        token_limit_input=200_000, token_limit_output=32_000
+                        # Input limit is 200_000, we leave 20_000 tokens as buffer due to tokenizer mismatch
+                        token_limit_input=180_000,
+                        token_limit_output=32_000,
                     ),
                     info_cutoff_at=date(2025, 3, 1),
                     published_at=date(2025, 5, 1),

{unique_toolkit-0.8.8.dist-info → unique_toolkit-0.8.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: unique_toolkit
-Version: 0.8.8
+Version: 0.8.9
 Summary: 
 License: Proprietary
 Author: Martin Fadler
@@ -114,6 +114,9 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), 
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.8.9] - 2025-08-15
+- Reduce input token limits for `ANTHROPIC_CLAUDE_3_7_SONNET_THINKING`, `ANTHROPIC_CLAUDE_3_7_SONNET`, `ANTHROPIC_CLAUDE_OPUS_4` and `ANTHROPIC_CLAUDE_SONNET_4` to 180_000 from 200_000
+
 ## [0.8.8] - 2025-08-11
 - Make chat service openai stream response openai compatible 
 - Make `ChatMessage` openai compatible

{unique_toolkit-0.8.8.dist-info → unique_toolkit-0.8.9.dist-info}/RECORD

@@ -55,7 +55,7 @@ unique_toolkit/language_model/__init__.py,sha256=lRQyLlbwHbNFf4-0foBU13UGb09lwEe
 unique_toolkit/language_model/builder.py,sha256=4OKfwJfj3TrgO1ezc_ewIue6W7BCQ2ZYQXUckWVPPTA,3369
 unique_toolkit/language_model/constants.py,sha256=B-topqW0r83dkC_25DeQfnPk3n53qzIHUCBS7YJ0-1U,119
 unique_toolkit/language_model/functions.py,sha256=qi4KGa4hLUXy0KwkkGmtGBM1IuYfJqmTheSQ95R7pNw,15611
-unique_toolkit/language_model/infos.py,sha256=mnUnbjDQNOIuPS2VE1SsgyKOxdRulo-9Z5k7_S1Q8Cw,37631
+unique_toolkit/language_model/infos.py,sha256=0Dq1ywdKLvtdLhkPycStt-DU_B5eito7EcJbzLLD1kU,38167
 unique_toolkit/language_model/prompt.py,sha256=JSawaLjQg3VR-E2fK8engFyJnNdk21zaO8pPIodzN4Q,3991
 unique_toolkit/language_model/reference.py,sha256=nkX2VFz-IrUz8yqyc3G5jUMNwrNpxITBrMEKkbqqYoI,8583
 unique_toolkit/language_model/schemas.py,sha256=p95cAwrYmEBupI8f7JZY17aJTKEHvtqF88hMIzS5r_k,16259
@@ -82,7 +82,7 @@ unique_toolkit/tools/utils/execution/execution.py,sha256=vjG2Y6awsGNtlvyQAGCTthQ
 unique_toolkit/tools/utils/source_handling/schema.py,sha256=pvNhtL2daDLpCVIQpfdn6R35GvKmITVLXjZNLAwpgUE,871
 unique_toolkit/tools/utils/source_handling/source_formatting.py,sha256=C7uayNbdkNVJdEARA5CENnHtNY1SU6etlaqbgHNyxaQ,9152
 unique_toolkit/tools/utils/source_handling/tests/test_source_formatting.py,sha256=zu3AJnYH9CMqZPrxKEH3IgI-fM3nlvIBuspJG6W6B18,6978
-unique_toolkit-0.8.8.dist-info/LICENSE,sha256=GlN8wHNdh53xwOPg44URnwag6TEolCjoq3YD_KrWgss,193
-unique_toolkit-0.8.8.dist-info/METADATA,sha256=AUly1ya9VxSfPCq8ivY1kAq4MrClSRQ-fm8_9BZr0oI,26649
-unique_toolkit-0.8.8.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-unique_toolkit-0.8.8.dist-info/RECORD,,
+unique_toolkit-0.8.9.dist-info/LICENSE,sha256=GlN8wHNdh53xwOPg44URnwag6TEolCjoq3YD_KrWgss,193
+unique_toolkit-0.8.9.dist-info/METADATA,sha256=GXgO-pTEsJdHBTjG3a_liQz8HddNZxTl19a-Zes5GE0,26859
+unique_toolkit-0.8.9.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+unique_toolkit-0.8.9.dist-info/RECORD,,