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 +1,35 @@
+name = "MiniMax-M2.5"
+description = "MiniMax model for chat, coding, office work, and agentic tasks"
+# Reasoning HTTP format (accessed 2026-06-25):
+# Coding Plan exact-string allowlist model; Alibaba-deployed MiniMax-M2.5 is
+# thinking-only. No toggle, effort, or numeric budget request field is documented.
+# Sources:
+# https://www.alibabacloud.com/help/en/model-studio/coding-plan
+# https://www.alibabacloud.com/help/en/model-studio/deep-thinking
+family = "minimax"
+release_date = "2026-02-12"
+last_updated = "2026-02-12"
+attachment = false
+reasoning = true
+reasoning_options = []
+temperature = true
+tool_call = true
+open_weights = true
+
+[interleaved]
+field = "reasoning_content"
+
+[cost]
+input = 0
+output = 0
+cache_read = 0
+cache_write = 0
+
+[limit]
+context = 196_608
+input = 196_601
+output = 24_576
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,29 @@
+name = "GLM-4.7"
+description = "Flagship GLM model for hybrid reasoning, coding, and agentic engineering"
+family = "glm"
+release_date = "2025-12-22"
+last_updated = "2025-12-22"
+attachment = false
+reasoning = true
+reasoning_options = [{ type = "toggle" }]
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = true
+
+[interleaved]
+field = "reasoning_content"
+
+[cost]
+input = 0
+output = 0
+cache_read = 0
+cache_write = 0
+
+[limit]
+context = 202_752
+output = 16_384
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,34 @@
+name = "GLM-5"
+description = "Flagship GLM model for hybrid reasoning, coding, and agentic engineering"
+# Reasoning HTTP format (accessed 2026-06-25):
+# Coding Plan exact-string allowlist model. Hybrid thinking is on by default and
+# top-level `enable_thinking`: true|false toggles it. No effort/budget documented.
+# Sources:
+# https://www.alibabacloud.com/help/en/model-studio/coding-plan
+# https://www.alibabacloud.com/help/en/model-studio/deep-thinking
+family = "glm"
+release_date = "2026-02-11"
+last_updated = "2026-02-11"
+attachment = false
+reasoning = true
+reasoning_options = [{ type = "toggle" }]
+temperature = true
+tool_call = true
+open_weights = false
+
+[interleaved]
+field = "reasoning_content"
+
+[cost]
+input = 0
+output = 0
+cache_read = 0
+cache_write = 0
+
+[limit]
+context = 202_752
+output = 16_384
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,36 @@
+name = "Kimi K2.5"
+description = "Kimi multimodal agent model for visual understanding, coding, and planning"
+# Reasoning HTTP format (accessed 2026-06-25):
+# Coding Plan exact-string allowlist model. Hybrid thinking uses top-level
+# `enable_thinking`: true enables; false (default) disables. No effort or numeric
+# budget bound is documented specifically for the plan endpoint.
+# Sources:
+# https://www.alibabacloud.com/help/en/model-studio/coding-plan
+# https://www.alibabacloud.com/help/en/model-studio/kimi-api
+family = "kimi-k2"
+release_date = "2026-01-27"
+last_updated = "2026-01-27"
+attachment = true
+reasoning = true
+reasoning_options = [{ type = "toggle" }]
+temperature = true
+tool_call = true
+knowledge = "2025-01"
+open_weights = true
+
+[interleaved]
+field = "reasoning_content"
+
+[cost]
+input = 0
+output = 0
+cache_read = 0
+cache_write = 0
+
+[limit]
+context = 262_144
+output = 32_768
+
+[modalities]
+input = ["text", "image", "video"]
+output = ["text"]
@@ -0,0 +1,25 @@
+name = "Qwen3 Coder Next"
+description = "Qwen coding model for software agents, repository edits, and code reasoning"
+family = "qwen"
+release_date = "2026-02-03"
+last_updated = "2026-02-03"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[cost]
+input = 0
+output = 0
+cache_read = 0
+cache_write = 0
+
+[limit]
+context = 262_144
+output = 65_536
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,30 @@
+name = "Qwen3 Coder Plus"
+description = "Qwen coding model for software agents, repository edits, and code reasoning"
+# Reasoning HTTP format (accessed 2026-06-25):
+# Coding Plan exact-string allowlist model. The plan docs document no reasoning
+# toggle, effort, or numeric budget field for this model.
+# Sources:
+# https://www.alibabacloud.com/help/en/model-studio/coding-plan
+family = "qwen"
+release_date = "2025-07-23"
+last_updated = "2025-07-23"
+attachment = false
+reasoning = false
+temperature = true
+knowledge = "2025-04"
+tool_call = true
+open_weights = true
+
+[cost]
+input = 0
+output = 0
+cache_read = 0
+cache_write = 0
+
+[limit]
+context = 1_000_000
+output = 65_536
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,25 @@
+name = "Qwen3 Max"
+description = "Flagship Qwen model for complex reasoning, coding, and agentic workflows"
+family = "qwen"
+release_date = "2026-01-23"
+last_updated = "2026-01-23"
+attachment = false
+reasoning = false
+temperature = true
+knowledge = "2025-04"
+tool_call = true
+open_weights = false
+
+[cost]
+input = 0
+output = 0
+cache_read = 0
+cache_write = 0
+
+[limit]
+context = 262_144
+output = 32_768
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,26 @@
+name = "Qwen3.5 Plus"
+description = "Qwen vision-language model for visual reasoning, documents, and agent tasks"
+family = "qwen"
+release_date = "2026-02-16"
+last_updated = "2026-02-16"
+attachment = false
+reasoning = true
+reasoning_options = [{ type = "toggle" }]
+temperature = true
+knowledge = "2025-04"
+tool_call = true
+open_weights = false
+
+[cost]
+input = 0
+output = 0
+cache_read = 0
+cache_write = 0
+
+[limit]
+context = 1_000_000
+output = 65_536
+
+[modalities]
+input = ["text", "image", "video"]
+output = ["text"]
@@ -0,0 +1,7 @@
+base_model = "alibaba/qwen3.6-flash"
+reasoning_options = [{ type = "toggle" }]
+
+[cost]
+input = 0.1875
+output = 1.125
+cache_write = 0.234375
@@ -0,0 +1,26 @@
+name = "Qwen3.6 Plus"
+description = "Qwen vision-language model for visual reasoning, documents, and agent tasks"
+family = "qwen"
+release_date = "2026-04-02"
+last_updated = "2026-04-02"
+attachment = false
+reasoning = true
+reasoning_options = [{ type = "toggle" }]
+temperature = true
+knowledge = "2025-04"
+tool_call = true
+open_weights = false
+
+[cost]
+input = 0
+output = 0
+cache_read = 0
+cache_write = 0
+
+[limit]
+context = 1_000_000
+output = 65_536
+
+[modalities]
+input = ["text", "image", "video"]
+output = ["text"]
@@ -0,0 +1,8 @@
+base_model = "alibaba/qwen3.7-max"
+reasoning_options = [{ type = "toggle" }]
+
+[cost]
+input = 2.5
+output = 7.5
+cache_read = 0.5
+cache_write = 3.125
@@ -0,0 +1,15 @@
+base_model = "alibaba/qwen3.7-plus"
+# Reasoning HTTP format (accessed 2026-06-25):
+# Coding Plan exact-string allowlist model. Hybrid thinking is on by default;
+# top-level `enable_thinking`: true|false toggles it. The plan docs do not state
+# a numeric budget bound or a plan-specific effort control.
+# Sources:
+# https://www.alibabacloud.com/help/en/model-studio/coding-plan
+# https://www.alibabacloud.com/help/en/model-studio/deep-thinking
+reasoning_options = [{ type = "toggle" }]
+
+[cost]
+input = 0
+output = 0
+cache_read = 0
+cache_write = 0