Update GreenPT: cached-token rates, compression variants, kimi-k3

providers/greenpt/models/deepseek-v4-flash-0731.toml

@@ -0,0 +1,16 @@
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
+# Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
+base_model = "deepseek/deepseek-v4-flash"
+name = "DeepSeek V4 Flash 0731"
+release_date = "2026-07-31"
+last_updated = "2026-07-31"
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 0.1596
+cache_read = 0.0456
+output = 0.399

providers/greenpt/models/gemma4.toml

@@ -1,7 +1,7 @@
 # Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
-# reasoning_effort: thinking is enabled by default, "none" disables it, and
-# minimal/low/medium/high are accepted. See https://docs.greenpt.ai/chat-completion
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
 base_model = "google/gemma-4-26b-a4b-it"
 name = "gemma4"
 reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]

providers/greenpt/models/glm-5.1.toml

@@ -1,6 +1,7 @@
 # Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
 base_model = "zhipuai/glm-5.1"
+status = "deprecated"
 reasoning_options = []
 
 [cost]

providers/greenpt/models/glm-5.2-caveman-lite.toml

@@ -0,0 +1,15 @@
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
+# Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
+base_model = "zhipuai/glm-5.2"
+name = "GLM-5.2 Caveman Lite"
+description = "glm-5.2 carrying a built-in ruleset that compresses prose, keeping code and technical detail verbatim. The gentlest tier: it cuts filler only and keeps the explanation intact. Same upstream model and price per token as glm-5.2, with fewer output tokens."
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 1.254
+cache_read = 0.3135
+output = 5.016

providers/greenpt/models/glm-5.2-caveman-ultra.toml

@@ -0,0 +1,15 @@
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
+# Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
+base_model = "zhipuai/glm-5.2"
+name = "GLM-5.2 Caveman Ultra"
+description = "glm-5.2 carrying a built-in ruleset that compresses prose, keeping code and technical detail verbatim. The most aggressive tier, close to answer-only. Same upstream model and price per token as glm-5.2, with fewer output tokens."
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 1.254
+cache_read = 0.3135
+output = 5.016

providers/greenpt/models/glm-5.2-caveman.toml

@@ -0,0 +1,15 @@
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
+# Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
+base_model = "zhipuai/glm-5.2"
+name = "GLM-5.2 Caveman"
+description = "glm-5.2 carrying a built-in ruleset that compresses prose, keeping code and technical detail verbatim. The middle tier, and the ruleset as its authors wrote it. Same upstream model and price per token as glm-5.2, with fewer output tokens."
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 1.254
+cache_read = 0.3135
+output = 5.016

providers/greenpt/models/glm-5.2-honey-lite.toml

@@ -0,0 +1,15 @@
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
+# Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
+base_model = "zhipuai/glm-5.2"
+name = "GLM-5.2 Honey Lite"
+description = "glm-5.2 carrying a built-in ruleset that compresses both generated code and prose. The gentlest tier: it cuts filler only and keeps the explanation intact. Same upstream model and price per token as glm-5.2, with fewer output tokens."
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 1.254
+cache_read = 0.3135
+output = 5.016

providers/greenpt/models/glm-5.2-honey-ultra.toml

@@ -0,0 +1,15 @@
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
+# Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
+base_model = "zhipuai/glm-5.2"
+name = "GLM-5.2 Honey Ultra"
+description = "glm-5.2 carrying a built-in ruleset that compresses both generated code and prose. The most aggressive tier, close to answer-only. Same upstream model and price per token as glm-5.2, with fewer output tokens."
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 1.254
+cache_read = 0.3135
+output = 5.016

providers/greenpt/models/glm-5.2-honey.toml

@@ -0,0 +1,15 @@
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
+# Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
+base_model = "zhipuai/glm-5.2"
+name = "GLM-5.2 Honey"
+description = "glm-5.2 carrying a built-in ruleset that compresses both generated code and prose. The middle tier, and the ruleset as its authors wrote it. Same upstream model and price per token as glm-5.2, with fewer output tokens."
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 1.254
+cache_read = 0.3135
+output = 5.016

providers/greenpt/models/glm-5.2-ponytail-lite.toml

@@ -0,0 +1,15 @@
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
+# Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
+base_model = "zhipuai/glm-5.2"
+name = "GLM-5.2 Ponytail Lite"
+description = "glm-5.2 carrying a built-in ruleset that compresses generated code, preferring platform features over custom code. The gentlest tier: it cuts filler only and keeps the explanation intact. Same upstream model and price per token as glm-5.2, with fewer output tokens."
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 1.254
+cache_read = 0.3135
+output = 5.016

providers/greenpt/models/glm-5.2-ponytail-ultra.toml

@@ -0,0 +1,15 @@
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
+# Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
+base_model = "zhipuai/glm-5.2"
+name = "GLM-5.2 Ponytail Ultra"
+description = "glm-5.2 carrying a built-in ruleset that compresses generated code, preferring platform features over custom code. The most aggressive tier, close to answer-only. Same upstream model and price per token as glm-5.2, with fewer output tokens."
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 1.254
+cache_read = 0.3135
+output = 5.016

providers/greenpt/models/glm-5.2-ponytail.toml

@@ -0,0 +1,15 @@
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
+# Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
+base_model = "zhipuai/glm-5.2"
+name = "GLM-5.2 Ponytail"
+description = "glm-5.2 carrying a built-in ruleset that compresses generated code, preferring platform features over custom code. The middle tier, and the ruleset as its authors wrote it. Same upstream model and price per token as glm-5.2, with fewer output tokens."
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 1.254
+cache_read = 0.3135
+output = 5.016

providers/greenpt/models/glm-5.2.toml

@@ -1,8 +1,13 @@
-# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
 base_model = "zhipuai/glm-5.2"
-reasoning_options = []
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
 
 [cost]
 input = 1.254
+cache_read = 0.3135
 output = 5.016

providers/greenpt/models/gpt-oss-120b.toml

@@ -1,8 +1,11 @@
 # Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: this endpoint accepts only the values below; the remaining
+# documented values are rejected with a 400. Verified against the production API.
+# See https://docs.greenpt.ai/chat-completion
 base_model = "openai/gpt-oss-120b"
 attachment = true
-reasoning_options = []
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
 
 [cost]
 input = 0.228

providers/greenpt/models/green-r-raw.toml

@@ -1,11 +1,12 @@
 # Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
-# reasoning_effort: thinking is enabled by default, "none" disables it, and
-# minimal/low/medium/high are accepted. See https://docs.greenpt.ai/chat-completion
+# reasoning_effort: this endpoint accepts only the values below; the remaining
+# documented values are rejected with a 400. Verified against the production API.
+# See https://docs.greenpt.ai/chat-completion
 base_model = "openai/gpt-oss-120b"
 name = "Green R Raw"
 attachment = true
-reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
 
 [cost]
 input = 0.399

providers/greenpt/models/green-r.toml

@@ -1,11 +1,12 @@
 # Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
-# reasoning_effort: thinking is enabled by default, "none" disables it, and
-# minimal/low/medium/high are accepted. See https://docs.greenpt.ai/chat-completion
+# reasoning_effort: this endpoint accepts only the values below; the remaining
+# documented values are rejected with a 400. Verified against the production API.
+# See https://docs.greenpt.ai/chat-completion
 base_model = "openai/gpt-oss-120b"
 name = "Green R"
 attachment = true
-reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
 
 [cost]
 input = 0.399

providers/greenpt/models/holo2-30b-a3b.toml

@@ -6,7 +6,7 @@ release_date = "2025-11"
 last_updated = "2025-11"
 attachment = true
 reasoning = true
-reasoning_options = []
+reasoning_options = [{ type = "effort", values = ["low", "medium", "high"] }]
 temperature = true
 tool_call = true
 structured_output = false

providers/greenpt/models/kimi-k2.6-fast.toml

@@ -2,6 +2,7 @@
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
 base_model = "moonshotai/kimi-k2.6"
 name = "Kimi K2.6 Fast"
+status = "deprecated"
 reasoning_options = []
 
 [cost]

providers/greenpt/models/kimi-k2.6.toml

@@ -1,11 +1,16 @@
-# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
 base_model = "moonshotai/kimi-k2.6"
-reasoning_options = []
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
 
 [cost]
-input = 0.828
-output = 4.389
+input = 0.7524
+cache_read = 0.2508
+output = 4.275
 
 [modalities]
 input = ["text", "image"]

providers/greenpt/models/kimi-k2.7-code.toml

@@ -1,11 +1,16 @@
-# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
 base_model = "moonshotai/kimi-k2.7-code"
 attachment = false
-reasoning_options = []
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
 
 [cost]
-input = 0.941
+input = 0.9006
+cache_read = 0.1881
 output = 4.389
 
 [modalities]

providers/greenpt/models/kimi-k3.toml

@@ -0,0 +1,16 @@
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
+# Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
+base_model = "moonshotai/kimi-k3"
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
+
+[cost]
+input = 3.762
+cache_read = 0.9405
+output = 18.81
+
+[modalities]
+input = ["text", "image"]

providers/greenpt/models/minimax-m2.5.toml

@@ -1,8 +1,13 @@
-# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
+# Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-08-01).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
+# cache_read: discounted rate for cached prompt tokens. Cache writes are not
+# charged. See https://docs.greenpt.ai/prompt-caching
 base_model = "minimax/MiniMax-M2.5"
-reasoning_options = []
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
 
 [cost]
-input = 0.188
+input = 0.1938
+cache_read = 0.0627
 output = 1.129

providers/greenpt/models/mistral-medium-3.5-128b.toml

@@ -1,7 +1,10 @@
 # Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: this endpoint accepts only the values below; the remaining
+# documented values are rejected with a 400. Verified against the production API.
+# See https://docs.greenpt.ai/chat-completion
 base_model = "mistral/mistral-medium-2604"
-reasoning_options = []
+reasoning_options = [{ type = "effort", values = ["none", "high"] }]
 
 [cost]
 input = 2.052

providers/greenpt/models/qwen3.5-397b-a17b.toml

@@ -1,8 +1,10 @@
 # Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
 base_model = "alibaba/qwen3.5-397b-a17b"
 attachment = false
-reasoning_options = []
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
 
 [cost]
 input = 0.798

providers/greenpt/models/qwen3.6-35b-a3b.toml

@@ -1,7 +1,9 @@
 # Cost: converted from GreenPT's EUR list price at 1.14 USD/EUR (rate captured 2026-07-24).
 # Sources: https://docs.greenpt.ai/pricing and https://docs.greenpt.ai/model-cards
+# reasoning_effort: thinking is enabled by default and "none" disables it.
+# Values verified against the production API. See https://docs.greenpt.ai/chat-completion
 base_model = "alibaba/qwen3.6-35b-a3b"
-reasoning_options = []
+reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }]
 
 [cost]
 input = 0.342