chore: import upstream snapshot with attribution

2026-07-13 12:28:55 +08:00
commit db42b91b75
6397 changed files with 146012 additions and 0 deletions
@@ -0,0 +1,11 @@
+base_model = "anthropic/claude-haiku-4-5"
+# Databricks' reasoning-model request table documents no toggle, effort, or
+# token-budget control for this endpoint.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = []
+
+[cost]
+input = 1
+output = 5
+cache_read = 0.1
+cache_write = 1.25
@@ -0,0 +1,11 @@
+base_model = "anthropic/claude-opus-4-1"
+# Chat uses `thinking = { type = "enabled", budget_tokens = N }`; N >= 1024
+# and must be less than `max_tokens`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "budget_tokens", min = 1_024 }]
+
+[cost]
+input = 15
+output = 75
+cache_read = 1.5
+cache_write = 18.75
@@ -0,0 +1,11 @@
+base_model = "anthropic/claude-opus-4-5"
+# Chat uses `thinking = { type = "enabled", budget_tokens = N }`; N >= 1024
+# and must be less than `max_tokens`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "budget_tokens", min = 1_024 }]
+
+[cost]
+input = 5
+output = 25
+cache_read = 0.5
+cache_write = 6.25
@@ -0,0 +1,15 @@
+base_model = "anthropic/claude-opus-4-6"
+# Chat uses `thinking = { type = "enabled", budget_tokens = N }`; N >= 1024
+# and must be less than `max_tokens`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "budget_tokens", min = 1_024 }]
+
+[cost]
+input = 5
+output = 25
+cache_read = 0.5
+cache_write = 6.25
+
+[experimental.modes.fast]
+cost = { input = 30, output = 150, cache_read = 3, cache_write = 37.5 }
+provider = { body = { speed = "fast" }, headers = { anthropic-beta = "fast-mode-2026-02-01" } }
@@ -0,0 +1,15 @@
+base_model = "anthropic/claude-opus-4-7"
+# Chat uses `thinking = { type = "enabled", budget_tokens = N }`; N >= 1024
+# and must be less than `max_tokens`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "budget_tokens", min = 1_024 }]
+
+[cost]
+input = 5
+output = 25
+cache_read = 0.5
+cache_write = 6.25
+
+[experimental.modes.fast]
+cost = { input = 30, output = 150, cache_read = 3, cache_write = 37.5 }
+provider = { body = { speed = "fast" }, headers = { anthropic-beta = "fast-mode-2026-02-01" } }
@@ -0,0 +1,11 @@
+base_model = "anthropic/claude-sonnet-4-5"
+# Chat uses `thinking = { type = "enabled", budget_tokens = N }`; N >= 1024
+# and must be less than `max_tokens`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "budget_tokens", min = 1_024 }]
+
+[cost]
+input = 3
+output = 15
+cache_read = 0.3
+cache_write = 3.75
@@ -0,0 +1,11 @@
+base_model = "anthropic/claude-sonnet-4-6"
+# Chat uses `thinking = { type = "enabled", budget_tokens = N }`; N >= 1024
+# and must be less than `max_tokens`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "budget_tokens", min = 1_024 }]
+
+[cost]
+input = 3
+output = 15
+cache_read = 0.3
+cache_write = 3.75
@@ -0,0 +1,11 @@
+base_model = "anthropic/claude-sonnet-4-5-20250929"
+# Chat uses `thinking = { type = "enabled", budget_tokens = N }`; N >= 1024
+# and must be less than `max_tokens`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "budget_tokens", min = 1_024 }]
+
+[cost]
+input = 3
+output = 15
+cache_read = 0.3
+cache_write = 3.75
@@ -0,0 +1,11 @@
+base_model = "google/gemini-2.5-flash"
+# Chat uses `thinking.budget_tokens` 0..24576; 0 disables and -1 requests dynamic
+# thinking in native Gemini, but -1 is a sentinel rather than a numeric budget.
+# https://ai.google.dev/gemini-api/docs/thinking (accessed 2026-06-25)
+reasoning_options = [{ type = "budget_tokens", min = 0, max = 24_576 }]
+
+[cost]
+input = 0.3
+output = 2.5
+cache_read = 0.03
+input_audio = 1
@@ -0,0 +1,16 @@
+base_model = "google/gemini-2.5-pro"
+# Chat uses `thinking.budget_tokens` 128..32768; reasoning cannot be disabled;
+# native Gemini -1 requests dynamic thinking but is not a numeric budget.
+# https://ai.google.dev/gemini-api/docs/thinking (accessed 2026-06-25)
+reasoning_options = [{ type = "budget_tokens", min = 128, max = 32_768 }]
+
+[cost]
+input = 1.25
+output = 10
+cache_read = 0.125
+
+[[cost.tiers]]
+tier = { type = "context", size = 200_000 }
+input = 2.5
+output = 15
+cache_read = 0.25
@@ -0,0 +1,10 @@
+base_model = "google/gemini-3.1-flash-lite-preview"
+# Chat `reasoning_effort = low|medium|high`; `low` is the default.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+
+[cost]
+input = 0.25
+output = 1.5
+cache_read = 0.025
+input_audio = 0.5
@@ -0,0 +1,15 @@
+base_model = "google/gemini-3.1-pro-preview-customtools"
+# Chat `reasoning_effort = low|medium|high`; `low` is the default.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+
+[cost]
+input = 2
+output = 12
+cache_read = 0.2
+
+[[cost.tiers]]
+tier = { type = "context", size = 200_000 }
+input = 4
+output = 18
+cache_read = 0.4
@@ -0,0 +1,10 @@
+base_model = "google/gemini-3-flash-preview"
+# Chat `reasoning_effort = low|medium|high`; `low` is the default.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+
+[cost]
+input = 0.5
+output = 3
+cache_read = 0.05
+input_audio = 1
@@ -0,0 +1,15 @@
+base_model = "google/gemini-3-pro-preview"
+# Chat `reasoning_effort = low|medium|high`; `low` is the default.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+
+[cost]
+input = 2
+output = 12
+cache_read = 0.2
+
+[[cost.tiers]]
+tier = { type = "context", size = 200_000 }
+input = 4
+output = 18
+cache_read = 0.4
@@ -0,0 +1,9 @@
+base_model = "zhipuai/glm-5.2"
+# Chat `reasoning_effort = low|medium|high`; `medium` is the default.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-07-11)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+
+[cost]
+input = 1.4
+output = 4.4
+cache_read = 0.26
@@ -0,0 +1,9 @@
+base_model = "openai/gpt-5.1"
+# Chat `reasoning_effort = none|low|medium|high`; `none` is the default.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["none", "low", "medium", "high"] }]
+
+[cost]
+input = 1.25
+output = 10
+cache_read = 0.125
@@ -0,0 +1,9 @@
+base_model = "openai/gpt-5.2"
+# Chat `reasoning_effort = none|low|medium|high`; `none` is the default.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["none", "low", "medium", "high"] }]
+
+[cost]
+input = 1.75
+output = 14
+cache_read = 0.175
@@ -0,0 +1,13 @@
+base_model = "openai/gpt-5.4-mini"
+# Chat `reasoning_effort = low|medium|high`; Responses uses `reasoning.effort`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+
+[cost]
+input = 0.75
+output = 4.5
+cache_read = 0.075
+
+[experimental.modes.fast]
+cost = { input = 1.5, output = 9, cache_read = 0.15 }
+provider = { body = { service_tier = "priority" } }
@@ -0,0 +1,9 @@
+base_model = "openai/gpt-5.4-nano"
+# Chat `reasoning_effort = low|medium|high`; Responses uses `reasoning.effort`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+
+[cost]
+input = 0.2
+output = 1.25
+cache_read = 0.02
@@ -0,0 +1,19 @@
+base_model = "openai/gpt-5.4"
+# Chat `reasoning_effort = low|medium|high`; Responses uses `reasoning.effort`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+
+[cost]
+input = 2.5
+output = 15
+cache_read = 0.25
+
+[[cost.tiers]]
+tier = { type = "context", size = 272_000 }
+input = 5
+output = 22.5
+cache_read = 0.5
+
+[experimental.modes.fast]
+cost = { input = 5, output = 30, cache_read = 0.5 }
+provider = { body = { service_tier = "priority" } }
@@ -0,0 +1,19 @@
+base_model = "openai/gpt-5.5"
+# Chat `reasoning_effort = low|medium|high`; Responses uses `reasoning.effort`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+
+[cost]
+input = 5
+output = 30
+cache_read = 0.5
+
+[[cost.tiers]]
+tier = { type = "context", size = 272_000 }
+input = 10
+output = 45
+cache_read = 1
+
+[experimental.modes.fast]
+cost = { input = 12.5, output = 75, cache_read = 1.25 }
+provider = { body = { service_tier = "priority" } }
@@ -0,0 +1,9 @@
+base_model = "openai/gpt-5-mini"
+# Chat `reasoning_effort = minimal|low|medium|high`; Responses uses `reasoning.effort`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 0.25
+output = 2
+cache_read = 0.025
@@ -0,0 +1,9 @@
+base_model = "openai/gpt-5-nano"
+# Chat `reasoning_effort = minimal|low|medium|high`; Responses uses `reasoning.effort`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 0.05
+output = 0.4
+cache_read = 0.005
@@ -0,0 +1,9 @@
+base_model = "openai/gpt-5"
+# Chat `reasoning_effort = minimal|low|medium|high`; Responses uses `reasoning.effort`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 1.25
+output = 10
+cache_read = 0.125
@@ -0,0 +1,26 @@
+name = "GPT OSS 120B"
+description = "Open-weight GPT model for self-hosted reasoning and instruction-following workloads"
+family = "gpt-oss"
+release_date = "2025-08-05"
+last_updated = "2025-08-05"
+attachment = false
+reasoning = true
+# Chat `reasoning_effort = low|medium|high`; `medium` is the default.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[cost]
+input = 0.072
+output = 0.28
+
+[limit]
+context = 131_072
+output = 32_768
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,26 @@
+name = "GPT OSS 20B"
+description = "Open-weight GPT model for self-hosted reasoning and instruction-following workloads"
+family = "gpt-oss"
+release_date = "2025-08-05"
+last_updated = "2025-08-05"
+attachment = false
+reasoning = true
+# Chat `reasoning_effort = low|medium|high`; `medium` is the default.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[cost]
+input = 0.05
+output = 0.20
+
+[limit]
+context = 131_072
+output = 32_768
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,9 @@
+base_model = "moonshotai/kimi-k2.7-code"
+# Chat `reasoning_effort = low|medium|high`; `medium` is the default.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-07-11)
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
+
+[cost]
+input = 0.95
+output = 4.0
+cache_read = 0.19
@@ -0,0 +1,15 @@
+name = "Databricks"
+npm = "@ai-sdk/openai-compatible"
+# Raw Chat is POST `/serving-endpoints/{model}/invocations`: GPT/Gemini 3 use
+# `reasoning_effort`; Claude/Gemini 2.5 use
+# `thinking = { type = "enabled", budget_tokens = N }`.
+# Native OpenAI Responses is POST `/serving-endpoints/responses` and uses
+# `reasoning.effort`; Open Responses supports Claude, Gemini, and open models.
+# Native Gemini `generateContent` uses `generationConfig.thinkingConfig`.
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-reason-models (accessed 2026-06-25)
+# https://docs.databricks.com/aws/en/machine-learning/foundation-model-apis/api-reference (accessed 2026-06-25)
+# https://docs.databricks.com/aws/en/machine-learning/model-serving/query-open-responses-models (accessed 2026-06-25)
+# https://docs.databricks.com/aws/en/generative-ai/external-models/google-gemini (accessed 2026-06-25)
+api = "https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1"
+env = ["DATABRICKS_HOST", "DATABRICKS_TOKEN"]
+doc = "https://docs.databricks.com/aws/en/machine-learning/foundation-models/"