unique-toolkit 2026.38.0.dev13__tar.gz → 2026.38.0.dev15__tar.gz

{unique_toolkit-2026.38.0.dev13 → unique_toolkit-2026.38.0.dev15}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: unique-toolkit
-Version: 2026.38.0.dev13
+Version: 2026.38.0.dev15
 Summary: 
 Author: Cedric Klinkert, Martin Fadler, Sadique Sheik, Fabian Schläpfer, Pascal Hauri, Ejiroghene Oki-Peter, Rami Azouz, Ricardo Visini, Sanzio Monti
 Author-email: Cedric Klinkert <cedric.klinkert@unique.ch>, Martin Fadler <martin.fadler@unique.ch>, Sadique Sheik <sadique@unique.ch>, Fabian Schläpfer <fabian@unique.ch>, Pascal Hauri <pascal@unique.ch>, Ejiroghene Oki-Peter <oki-peter.ejiroghene@unique.ai>, Rami Azouz <rami.ext@unique.ch>, Ricardo Visini <ricardo.visini@unique.ch>, Sanzio Monti <sanzio.monti@unique.ai>

{unique_toolkit-2026.38.0.dev13 → unique_toolkit-2026.38.0.dev15}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "unique_toolkit"
-version = "2026.38.0.dev13"
+version = "2026.38.0.dev15"
 description = ""
 readme = "README.md"
 requires-python = ">=3.12"

{unique_toolkit-2026.38.0.dev13 → unique_toolkit-2026.38.0.dev15}/unique_toolkit/language_model/infos.py

@@ -64,6 +64,7 @@ class LanguageModelName(StrEnum):
     AZURE_GPT_56_SOL_2026_0709 = "AZURE_GPT_56_SOL_2026_0709"
     AZURE_GPT_56_TERRA_2026_0709 = "AZURE_GPT_56_TERRA_2026_0709"
     AZURE_GPT_56_LUNA_2026_0709 = "AZURE_GPT_56_LUNA_2026_0709"
+    AZURE_GPT_6_ASTRA_2026_0903 = "AZURE_GPT_6_ASTRA_2026_0903"
     AZURE_GPT_4o_2024_0513 = "AZURE_GPT_4o_2024_0513"
     AZURE_GPT_4o_2024_0806 = "AZURE_GPT_4o_2024_0806"
     AZURE_GPT_4o_2024_1120 = "AZURE_GPT_4o_2024_1120"
@@ -127,6 +128,7 @@ class LanguageModelName(StrEnum):
     LITELLM_OPENAI_GPT_56_SOL = "litellm:openai-gpt-5-6-sol"
     LITELLM_OPENAI_GPT_56_TERRA = "litellm:openai-gpt-5-6-terra"
     LITELLM_OPENAI_GPT_56_LUNA = "litellm:openai-gpt-5-6-luna"
+    LITELLM_OPENAI_GPT_6_ASTRA = "litellm:openai-gpt-6-astra"
     LITELLM_OPENAI_O1 = "litellm:openai-o1"
     LITELLM_OPENAI_O3 = "litellm:openai-o3"
     LITELLM_OPENAI_O3_DEEP_RESEARCH = "litellm:openai-o3-deep-research"
@@ -219,6 +221,7 @@ def get_encoder_name(model_name: LanguageModelName) -> EncoderName:
             | LMN.AZURE_GPT_56_SOL_2026_0709
             | LMN.AZURE_GPT_56_TERRA_2026_0709
             | LMN.AZURE_GPT_56_LUNA_2026_0709
+            | LMN.AZURE_GPT_6_ASTRA_2026_0903
             | LMN.AZURE_MODEL_ROUTER_2025_1118
             | LMN.LITELLM_OPENAI_GPT_5
             | LMN.LITELLM_OPENAI_GPT_5_MINI
@@ -236,6 +239,7 @@ def get_encoder_name(model_name: LanguageModelName) -> EncoderName:
             | LMN.LITELLM_OPENAI_GPT_56_SOL
             | LMN.LITELLM_OPENAI_GPT_56_TERRA
             | LMN.LITELLM_OPENAI_GPT_56_LUNA
+            | LMN.LITELLM_OPENAI_GPT_6_ASTRA
             | LMN.LITELLM_OPENAI_O1
             | LMN.LITELLM_OPENAI_O3
             | LMN.LITELLM_OPENAI_O3_DEEP_RESEARCH
@@ -827,6 +831,12 @@ class LanguageModelInfo(BaseModel):
                     default_options={
                         "reasoning_effort": "none",
                     },
+                    # Do NOT add "xhigh": xhigh is Codex-only in the GPT-5.1
+                    # family. Verified on QA (switzerlandnorth, 2026-09-04):
+                    # the deployment accepts reasoning_effort=xhigh on the
+                    # wire but returns 0 reasoning tokens (vs ~2'200-2'400 at
+                    # "high"), i.e. it silently runs without reasoning.
+                    # See UN-25265.
                     supported_reasoning_efforts=["none", "low", "medium", "high"],
                 )
             case LanguageModelName.AZURE_GPT_51_THINKING_2025_1113:
@@ -889,6 +899,10 @@ class LanguageModelInfo(BaseModel):
                     },
                     supported_reasoning_efforts=["low", "medium", "high"],
                 )
+            # The GPT-5.1 Codex family supports reasoning_effort="xhigh"
+            # (unlike base GPT-5.1, see comment there / UN-25265). No Azure
+            # deployment of the Codex models exists yet (as of 2026-09), so
+            # the server side could not be verified end-to-end.
             case LanguageModelName.AZURE_GPT_51_CODEX_2025_1113:
                 return cls(
                     name=model_name,
@@ -916,7 +930,13 @@ class LanguageModelInfo(BaseModel):
                     default_options={
                         "reasoning_effort": "medium",
                     },
-                    supported_reasoning_efforts=["none", "low", "medium", "high"],
+                    supported_reasoning_efforts=[
+                        "none",
+                        "low",
+                        "medium",
+                        "high",
+                        "xhigh",
+                    ],
                 )
             case LanguageModelName.AZURE_GPT_51_CODEX_MINI_2025_1113:
                 return cls(
@@ -945,7 +965,13 @@ class LanguageModelInfo(BaseModel):
                     default_options={
                         "reasoning_effort": "medium",
                     },
-                    supported_reasoning_efforts=["none", "low", "medium", "high"],
+                    supported_reasoning_efforts=[
+                        "none",
+                        "low",
+                        "medium",
+                        "high",
+                        "xhigh",
+                    ],
                 )
             case LanguageModelName.AZURE_GPT_52_2025_1211:
                 return cls(
@@ -1183,6 +1209,41 @@ class LanguageModelInfo(BaseModel):
                         "xhigh",
                     ],
                 )
+            case LanguageModelName.AZURE_GPT_6_ASTRA_2026_0903:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.AZURE,
+                    family=ModelFamily.OPENAI,
+                    version="2026-09-03",
+                    encoder_name=EncoderName.O200K_BASE,
+                    capabilities=[
+                        ModelCapabilities.CHAT_COMPLETIONS_API,
+                        ModelCapabilities.FUNCTION_CALLING,
+                        ModelCapabilities.PARALLEL_FUNCTION_CALLING,
+                        ModelCapabilities.REASONING,
+                        ModelCapabilities.RESPONSES_API,
+                        ModelCapabilities.STREAMING,
+                        ModelCapabilities.STRUCTURED_OUTPUT,
+                        ModelCapabilities.VISION,
+                    ],
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=922_000, token_limit_output=128_000
+                    ),
+                    info_cutoff_at=date(2026, 4, 30),
+                    published_at=date(2026, 9, 3),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=0.0, max_temperature=1.0
+                    ),
+                    default_options={
+                        "reasoning_effort": "low",
+                    },
+                    supported_reasoning_efforts=[
+                        "low",
+                        "medium",
+                        "high",
+                        "xhigh",
+                    ],
+                )
             case LanguageModelName.AZURE_GPT_4_TURBO_2024_0409:
                 return cls(
                     name=model_name,
@@ -2665,6 +2726,41 @@ class LanguageModelInfo(BaseModel):
                         "xhigh",
                     ],
                 )
+            case LanguageModelName.LITELLM_OPENAI_GPT_6_ASTRA:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.LITELLM,
+                    family=ModelFamily.OPENAI,
+                    version="2026-09-03",
+                    encoder_name=EncoderName.O200K_BASE,
+                    capabilities=[
+                        ModelCapabilities.CHAT_COMPLETIONS_API,
+                        ModelCapabilities.FUNCTION_CALLING,
+                        ModelCapabilities.PARALLEL_FUNCTION_CALLING,
+                        ModelCapabilities.REASONING,
+                        ModelCapabilities.RESPONSES_API,
+                        ModelCapabilities.STREAMING,
+                        ModelCapabilities.STRUCTURED_OUTPUT,
+                        ModelCapabilities.VISION,
+                    ],
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=922_000, token_limit_output=128_000
+                    ),
+                    info_cutoff_at=date(2026, 4, 30),
+                    published_at=date(2026, 9, 3),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=0.0, max_temperature=1.0
+                    ),
+                    default_options={
+                        "reasoning_effort": "low",
+                    },
+                    supported_reasoning_efforts=[
+                        "low",
+                        "medium",
+                        "high",
+                        "xhigh",
+                    ],
+                )
             case LanguageModelName.LITELLM_OPENAI_O1:
                 return cls(
                     name=model_name,