From a2168dbf13ef9f08b6c187a8018eabf4ded8ec35 Mon Sep 17 00:00:00 2001 From: Cas Burggraaf Date: Sat, 1 Aug 2026 09:06:27 +0200 Subject: [PATCH 1/8] Publish GreenPT cached-token rates and refresh prices GreenPT now bills prompt-cache hits at a reduced input rate on these models, so each gains cost.cache_read. Cache writes are not charged, so cost.cache_write is omitted rather than set to zero. glm-5.2 cache_read 0.3135 kimi-k2.6 cache_read 0.2508 kimi-k2.7-code cache_read 0.1881 minimax-m2.5 cache_read 0.0627 The same pass also picks up list-price corrections: kimi-k2.6 moves to 0.7524 / 4.275, kimi-k2.7-code input to 0.9006, and minimax-m2.5 input to 0.1938. glm-5.2's own prices are unchanged. Rates: https://docs.greenpt.ai/prompt-caching and https://docs.greenpt.ai/pricing --- providers/greenpt/models/glm-5.2.toml | 5 ++++- providers/greenpt/models/kimi-k2.6.toml | 9 ++++++--- providers/greenpt/models/kimi-k2.7-code.toml | 7 +++++-- providers/greenpt/models/minimax-m2.5.toml | 7 +++++-- 4 files changed, 20 insertions(+), 8 deletions(-) diff --git a/providers/greenpt/models/glm-5.2.toml b/providers/greenpt/models/glm-5.2.toml index 46a73261ca1..96aad6bae5a 100644 --- a/providers/greenpt/models/glm-5.2.toml +++ b/providers/greenpt/models/glm-5.2.toml @@ -1,8 +1,11 @@ -# 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 +# 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 = [] [cost] input = 1.254 +cache_read = 0.3135 output = 5.016 diff --git a/providers/greenpt/models/kimi-k2.6.toml b/providers/greenpt/models/kimi-k2.6.toml index 61db90787da..9f62ad102f3 100644 --- a/providers/greenpt/models/kimi-k2.6.toml +++ b/providers/greenpt/models/kimi-k2.6.toml @@ -1,11 +1,14 @@ -# 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 +# 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 = [] [cost] -input = 0.828 -output = 4.389 +input = 0.7524 +cache_read = 0.2508 +output = 4.275 [modalities] input = ["text", "image"] diff --git a/providers/greenpt/models/kimi-k2.7-code.toml b/providers/greenpt/models/kimi-k2.7-code.toml index dc90ccaa23d..50c4c74b848 100644 --- a/providers/greenpt/models/kimi-k2.7-code.toml +++ b/providers/greenpt/models/kimi-k2.7-code.toml @@ -1,11 +1,14 @@ -# 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 +# 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 = [] [cost] -input = 0.941 +input = 0.9006 +cache_read = 0.1881 output = 4.389 [modalities] diff --git a/providers/greenpt/models/minimax-m2.5.toml b/providers/greenpt/models/minimax-m2.5.toml index b276e1959cc..74aba2b6da4 100644 --- a/providers/greenpt/models/minimax-m2.5.toml +++ b/providers/greenpt/models/minimax-m2.5.toml @@ -1,8 +1,11 @@ -# 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 +# 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 = [] [cost] -input = 0.188 +input = 0.1938 +cache_read = 0.0627 output = 1.129 From e024ac26f4f4ca6ccb6ad6eb79baa0ecc3cb14a4 Mon Sep 17 00:00:00 2001 From: Cas Burggraaf Date: Sat, 1 Aug 2026 09:06:27 +0200 Subject: [PATCH 2/8] Add kimi-k3 to GreenPT Kimi K3 is generally available on GreenPT at 3.762 input, 18.81 output and 0.9405 for cached prompt tokens. GreenPT serves it with text and image input, so the inherited video modality is overridden away. https://docs.greenpt.ai/model-cards --- providers/greenpt/models/kimi-k3.toml | 14 ++++++++++++++ 1 file changed, 14 insertions(+) create mode 100644 providers/greenpt/models/kimi-k3.toml diff --git a/providers/greenpt/models/kimi-k3.toml b/providers/greenpt/models/kimi-k3.toml new file mode 100644 index 00000000000..c26e5fce611 --- /dev/null +++ b/providers/greenpt/models/kimi-k3.toml @@ -0,0 +1,14 @@ +# 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 +# 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 = [] + +[cost] +input = 3.762 +cache_read = 0.9405 +output = 18.81 + +[modalities] +input = ["text", "image"] From 75416fc569006dc4b810ca404a0f8c62f987df41 Mon Sep 17 00:00:00 2001 From: Cas Burggraaf Date: Sat, 1 Aug 2026 09:06:28 +0200 Subject: [PATCH 3/8] Add the nine GreenPT glm-5.2 compression variants GreenPT serves nine ids that are glm-5.2 carrying a built-in output-compression ruleset: three families (caveman compresses prose, ponytail compresses generated code, honey compresses both) at three intensities (-lite, unsuffixed, -ultra). They are the same upstream model at the same price per token, including the same cached rate, and return fewer output tokens. Each is declared through base_model so cost and limits cannot drift from glm-5.2. https://docs.greenpt.ai/compression-models --- providers/greenpt/models/glm-5.2-caveman-lite.toml | 13 +++++++++++++ providers/greenpt/models/glm-5.2-caveman-ultra.toml | 13 +++++++++++++ providers/greenpt/models/glm-5.2-caveman.toml | 13 +++++++++++++ providers/greenpt/models/glm-5.2-honey-lite.toml | 13 +++++++++++++ providers/greenpt/models/glm-5.2-honey-ultra.toml | 13 +++++++++++++ providers/greenpt/models/glm-5.2-honey.toml | 13 +++++++++++++ providers/greenpt/models/glm-5.2-ponytail-lite.toml | 13 +++++++++++++ .../greenpt/models/glm-5.2-ponytail-ultra.toml | 13 +++++++++++++ providers/greenpt/models/glm-5.2-ponytail.toml | 13 +++++++++++++ 9 files changed, 117 insertions(+) create mode 100644 providers/greenpt/models/glm-5.2-caveman-lite.toml create mode 100644 providers/greenpt/models/glm-5.2-caveman-ultra.toml create mode 100644 providers/greenpt/models/glm-5.2-caveman.toml create mode 100644 providers/greenpt/models/glm-5.2-honey-lite.toml create mode 100644 providers/greenpt/models/glm-5.2-honey-ultra.toml create mode 100644 providers/greenpt/models/glm-5.2-honey.toml create mode 100644 providers/greenpt/models/glm-5.2-ponytail-lite.toml create mode 100644 providers/greenpt/models/glm-5.2-ponytail-ultra.toml create mode 100644 providers/greenpt/models/glm-5.2-ponytail.toml diff --git a/providers/greenpt/models/glm-5.2-caveman-lite.toml b/providers/greenpt/models/glm-5.2-caveman-lite.toml new file mode 100644 index 00000000000..d8160adf44f --- /dev/null +++ b/providers/greenpt/models/glm-5.2-caveman-lite.toml @@ -0,0 +1,13 @@ +# 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 +# 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 = [] + +[cost] +input = 1.254 +cache_read = 0.3135 +output = 5.016 diff --git a/providers/greenpt/models/glm-5.2-caveman-ultra.toml b/providers/greenpt/models/glm-5.2-caveman-ultra.toml new file mode 100644 index 00000000000..7ae11445e96 --- /dev/null +++ b/providers/greenpt/models/glm-5.2-caveman-ultra.toml @@ -0,0 +1,13 @@ +# 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 +# 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 = [] + +[cost] +input = 1.254 +cache_read = 0.3135 +output = 5.016 diff --git a/providers/greenpt/models/glm-5.2-caveman.toml b/providers/greenpt/models/glm-5.2-caveman.toml new file mode 100644 index 00000000000..eadf7067340 --- /dev/null +++ b/providers/greenpt/models/glm-5.2-caveman.toml @@ -0,0 +1,13 @@ +# 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 +# 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 = [] + +[cost] +input = 1.254 +cache_read = 0.3135 +output = 5.016 diff --git a/providers/greenpt/models/glm-5.2-honey-lite.toml b/providers/greenpt/models/glm-5.2-honey-lite.toml new file mode 100644 index 00000000000..d2f4e93a387 --- /dev/null +++ b/providers/greenpt/models/glm-5.2-honey-lite.toml @@ -0,0 +1,13 @@ +# 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 +# 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 = [] + +[cost] +input = 1.254 +cache_read = 0.3135 +output = 5.016 diff --git a/providers/greenpt/models/glm-5.2-honey-ultra.toml b/providers/greenpt/models/glm-5.2-honey-ultra.toml new file mode 100644 index 00000000000..35db8a9ad2b --- /dev/null +++ b/providers/greenpt/models/glm-5.2-honey-ultra.toml @@ -0,0 +1,13 @@ +# 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 +# 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 = [] + +[cost] +input = 1.254 +cache_read = 0.3135 +output = 5.016 diff --git a/providers/greenpt/models/glm-5.2-honey.toml b/providers/greenpt/models/glm-5.2-honey.toml new file mode 100644 index 00000000000..46fdd74e37a --- /dev/null +++ b/providers/greenpt/models/glm-5.2-honey.toml @@ -0,0 +1,13 @@ +# 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 +# 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 = [] + +[cost] +input = 1.254 +cache_read = 0.3135 +output = 5.016 diff --git a/providers/greenpt/models/glm-5.2-ponytail-lite.toml b/providers/greenpt/models/glm-5.2-ponytail-lite.toml new file mode 100644 index 00000000000..8397426e703 --- /dev/null +++ b/providers/greenpt/models/glm-5.2-ponytail-lite.toml @@ -0,0 +1,13 @@ +# 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 +# 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 = [] + +[cost] +input = 1.254 +cache_read = 0.3135 +output = 5.016 diff --git a/providers/greenpt/models/glm-5.2-ponytail-ultra.toml b/providers/greenpt/models/glm-5.2-ponytail-ultra.toml new file mode 100644 index 00000000000..f79b905334a --- /dev/null +++ b/providers/greenpt/models/glm-5.2-ponytail-ultra.toml @@ -0,0 +1,13 @@ +# 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 +# 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 = [] + +[cost] +input = 1.254 +cache_read = 0.3135 +output = 5.016 diff --git a/providers/greenpt/models/glm-5.2-ponytail.toml b/providers/greenpt/models/glm-5.2-ponytail.toml new file mode 100644 index 00000000000..7981b1f928e --- /dev/null +++ b/providers/greenpt/models/glm-5.2-ponytail.toml @@ -0,0 +1,13 @@ +# 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 +# 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 = [] + +[cost] +input = 1.254 +cache_read = 0.3135 +output = 5.016 From 59bea96f0e3c7467d1ef727b5bd2afff10119251 Mon Sep 17 00:00:00 2001 From: Cas Burggraaf Date: Sat, 1 Aug 2026 09:06:28 +0200 Subject: [PATCH 4/8] Mark GreenPT kimi-k2.6-fast as deprecated The upstream provider withdrew this model and GreenPT no longer serves the id, so requests for it now fail. Marked deprecated rather than deleted so existing configurations still resolve against the catalog. --- providers/greenpt/models/kimi-k2.6-fast.toml | 1 + 1 file changed, 1 insertion(+) diff --git a/providers/greenpt/models/kimi-k2.6-fast.toml b/providers/greenpt/models/kimi-k2.6-fast.toml index b2ceae547af..9719eba4424 100644 --- a/providers/greenpt/models/kimi-k2.6-fast.toml +++ b/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] From c7e19a6f38e8ff9494c7d34fe907c0a3a9c5f13f Mon Sep 17 00:00:00 2001 From: Cas Burggraaf Date: Tue, 4 Aug 2026 14:37:34 +0200 Subject: [PATCH 5/8] Mark GreenPT glm-5.1 as deprecated The id is still advertised by /v1/models but every request for it returns 404 from production, so it is not servable. Marked deprecated rather than deleted, matching how kimi-k2.6-fast is handled here. --- providers/greenpt/models/glm-5.1.toml | 1 + 1 file changed, 1 insertion(+) diff --git a/providers/greenpt/models/glm-5.1.toml b/providers/greenpt/models/glm-5.1.toml index 3761f5ac117..ac773d36c29 100644 --- a/providers/greenpt/models/glm-5.1.toml +++ b/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] From dece881ea07a196c48158231be887a93ee1f18cf Mon Sep 17 00:00:00 2001 From: Cas Burggraaf Date: Tue, 4 Aug 2026 14:37:35 +0200 Subject: [PATCH 6/8] Declare the reasoning_effort values each GreenPT model accepts Replaces the blanket reasoning_options = [] with the values each endpoint actually accepts, established by sending every documented value to every model on the production API. The sets are not uniform, which is why the previous blanket declaration was wrong in both directions: none, minimal, low, medium, high glm-5.2 and its nine compression variants, kimi-k3, kimi-k2.6, kimi-k2.7-code, minimax-m2.5, qwen3.5-397b, qwen3.6-35b, gemma4 low, medium, high green-r, green-r-raw, gpt-oss-120b, holo2-30b-a3b (none and minimal return 400) none, high mistral-medium-3.5-128b (minimal, low and medium return 400) This also corrects green-r and green-r-raw, which previously advertised none and minimal even though both are rejected. On glm-5.2 and its variants the control is observable, not just accepted: reasoning_effort "none" takes the reported reasoning tokens to zero. --- providers/greenpt/models/gemma4.toml | 4 ++-- providers/greenpt/models/glm-5.2-caveman-lite.toml | 4 +++- providers/greenpt/models/glm-5.2-caveman-ultra.toml | 4 +++- providers/greenpt/models/glm-5.2-caveman.toml | 4 +++- providers/greenpt/models/glm-5.2-honey-lite.toml | 4 +++- providers/greenpt/models/glm-5.2-honey-ultra.toml | 4 +++- providers/greenpt/models/glm-5.2-honey.toml | 4 +++- providers/greenpt/models/glm-5.2-ponytail-lite.toml | 4 +++- providers/greenpt/models/glm-5.2-ponytail-ultra.toml | 4 +++- providers/greenpt/models/glm-5.2-ponytail.toml | 4 +++- providers/greenpt/models/glm-5.2.toml | 4 +++- providers/greenpt/models/gpt-oss-120b.toml | 5 ++++- providers/greenpt/models/green-r-raw.toml | 7 ++++--- providers/greenpt/models/green-r.toml | 7 ++++--- providers/greenpt/models/holo2-30b-a3b.toml | 2 +- providers/greenpt/models/kimi-k2.6.toml | 4 +++- providers/greenpt/models/kimi-k2.7-code.toml | 4 +++- providers/greenpt/models/kimi-k3.toml | 4 +++- providers/greenpt/models/minimax-m2.5.toml | 4 +++- providers/greenpt/models/mistral-medium-3.5-128b.toml | 5 ++++- providers/greenpt/models/qwen3.5-397b-a17b.toml | 4 +++- providers/greenpt/models/qwen3.6-35b-a3b.toml | 4 +++- 22 files changed, 67 insertions(+), 27 deletions(-) diff --git a/providers/greenpt/models/gemma4.toml b/providers/greenpt/models/gemma4.toml index 6371ffb3261..db0ce7c7753 100644 --- a/providers/greenpt/models/gemma4.toml +++ b/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"] }] diff --git a/providers/greenpt/models/glm-5.2-caveman-lite.toml b/providers/greenpt/models/glm-5.2-caveman-lite.toml index d8160adf44f..4b2aecce007 100644 --- a/providers/greenpt/models/glm-5.2-caveman-lite.toml +++ b/providers/greenpt/models/glm-5.2-caveman-lite.toml @@ -1,11 +1,13 @@ # 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 = [] +reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }] [cost] input = 1.254 diff --git a/providers/greenpt/models/glm-5.2-caveman-ultra.toml b/providers/greenpt/models/glm-5.2-caveman-ultra.toml index 7ae11445e96..7335fb86870 100644 --- a/providers/greenpt/models/glm-5.2-caveman-ultra.toml +++ b/providers/greenpt/models/glm-5.2-caveman-ultra.toml @@ -1,11 +1,13 @@ # 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 = [] +reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }] [cost] input = 1.254 diff --git a/providers/greenpt/models/glm-5.2-caveman.toml b/providers/greenpt/models/glm-5.2-caveman.toml index eadf7067340..4b9bfa2f782 100644 --- a/providers/greenpt/models/glm-5.2-caveman.toml +++ b/providers/greenpt/models/glm-5.2-caveman.toml @@ -1,11 +1,13 @@ # 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 = [] +reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }] [cost] input = 1.254 diff --git a/providers/greenpt/models/glm-5.2-honey-lite.toml b/providers/greenpt/models/glm-5.2-honey-lite.toml index d2f4e93a387..16fde976b0e 100644 --- a/providers/greenpt/models/glm-5.2-honey-lite.toml +++ b/providers/greenpt/models/glm-5.2-honey-lite.toml @@ -1,11 +1,13 @@ # 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 = [] +reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }] [cost] input = 1.254 diff --git a/providers/greenpt/models/glm-5.2-honey-ultra.toml b/providers/greenpt/models/glm-5.2-honey-ultra.toml index 35db8a9ad2b..bac13181da3 100644 --- a/providers/greenpt/models/glm-5.2-honey-ultra.toml +++ b/providers/greenpt/models/glm-5.2-honey-ultra.toml @@ -1,11 +1,13 @@ # 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 = [] +reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }] [cost] input = 1.254 diff --git a/providers/greenpt/models/glm-5.2-honey.toml b/providers/greenpt/models/glm-5.2-honey.toml index 46fdd74e37a..2f9eca1fb5c 100644 --- a/providers/greenpt/models/glm-5.2-honey.toml +++ b/providers/greenpt/models/glm-5.2-honey.toml @@ -1,11 +1,13 @@ # 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 = [] +reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }] [cost] input = 1.254 diff --git a/providers/greenpt/models/glm-5.2-ponytail-lite.toml b/providers/greenpt/models/glm-5.2-ponytail-lite.toml index 8397426e703..a6c9fcaeef1 100644 --- a/providers/greenpt/models/glm-5.2-ponytail-lite.toml +++ b/providers/greenpt/models/glm-5.2-ponytail-lite.toml @@ -1,11 +1,13 @@ # 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 = [] +reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }] [cost] input = 1.254 diff --git a/providers/greenpt/models/glm-5.2-ponytail-ultra.toml b/providers/greenpt/models/glm-5.2-ponytail-ultra.toml index f79b905334a..4d89df769be 100644 --- a/providers/greenpt/models/glm-5.2-ponytail-ultra.toml +++ b/providers/greenpt/models/glm-5.2-ponytail-ultra.toml @@ -1,11 +1,13 @@ # 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 = [] +reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }] [cost] input = 1.254 diff --git a/providers/greenpt/models/glm-5.2-ponytail.toml b/providers/greenpt/models/glm-5.2-ponytail.toml index 7981b1f928e..efc83972838 100644 --- a/providers/greenpt/models/glm-5.2-ponytail.toml +++ b/providers/greenpt/models/glm-5.2-ponytail.toml @@ -1,11 +1,13 @@ # 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 = [] +reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }] [cost] input = 1.254 diff --git a/providers/greenpt/models/glm-5.2.toml b/providers/greenpt/models/glm-5.2.toml index 96aad6bae5a..643695608cc 100644 --- a/providers/greenpt/models/glm-5.2.toml +++ b/providers/greenpt/models/glm-5.2.toml @@ -1,9 +1,11 @@ # 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 diff --git a/providers/greenpt/models/gpt-oss-120b.toml b/providers/greenpt/models/gpt-oss-120b.toml index 50b5d5e9126..6eb5952cf96 100644 --- a/providers/greenpt/models/gpt-oss-120b.toml +++ b/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 diff --git a/providers/greenpt/models/green-r-raw.toml b/providers/greenpt/models/green-r-raw.toml index 7dae641892c..eb4349b7745 100644 --- a/providers/greenpt/models/green-r-raw.toml +++ b/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 diff --git a/providers/greenpt/models/green-r.toml b/providers/greenpt/models/green-r.toml index 64969fede52..6b0e6568dbb 100644 --- a/providers/greenpt/models/green-r.toml +++ b/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 diff --git a/providers/greenpt/models/holo2-30b-a3b.toml b/providers/greenpt/models/holo2-30b-a3b.toml index 9ed8a828658..95c94bbdfab 100644 --- a/providers/greenpt/models/holo2-30b-a3b.toml +++ b/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 diff --git a/providers/greenpt/models/kimi-k2.6.toml b/providers/greenpt/models/kimi-k2.6.toml index 9f62ad102f3..d44ba86f556 100644 --- a/providers/greenpt/models/kimi-k2.6.toml +++ b/providers/greenpt/models/kimi-k2.6.toml @@ -1,9 +1,11 @@ # 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 = 0.7524 diff --git a/providers/greenpt/models/kimi-k2.7-code.toml b/providers/greenpt/models/kimi-k2.7-code.toml index 50c4c74b848..9e8e1f705ce 100644 --- a/providers/greenpt/models/kimi-k2.7-code.toml +++ b/providers/greenpt/models/kimi-k2.7-code.toml @@ -1,10 +1,12 @@ # 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 = 0.9006 diff --git a/providers/greenpt/models/kimi-k3.toml b/providers/greenpt/models/kimi-k3.toml index c26e5fce611..e6a6e4d9c35 100644 --- a/providers/greenpt/models/kimi-k3.toml +++ b/providers/greenpt/models/kimi-k3.toml @@ -1,9 +1,11 @@ # 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 = [] +reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }] [cost] input = 3.762 diff --git a/providers/greenpt/models/minimax-m2.5.toml b/providers/greenpt/models/minimax-m2.5.toml index 74aba2b6da4..d9b7de3be95 100644 --- a/providers/greenpt/models/minimax-m2.5.toml +++ b/providers/greenpt/models/minimax-m2.5.toml @@ -1,9 +1,11 @@ # 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 = 0.1938 diff --git a/providers/greenpt/models/mistral-medium-3.5-128b.toml b/providers/greenpt/models/mistral-medium-3.5-128b.toml index 83fa798e132..87e1acb7854 100644 --- a/providers/greenpt/models/mistral-medium-3.5-128b.toml +++ b/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 diff --git a/providers/greenpt/models/qwen3.5-397b-a17b.toml b/providers/greenpt/models/qwen3.5-397b-a17b.toml index 345573c6b70..302d942ca68 100644 --- a/providers/greenpt/models/qwen3.5-397b-a17b.toml +++ b/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 diff --git a/providers/greenpt/models/qwen3.6-35b-a3b.toml b/providers/greenpt/models/qwen3.6-35b-a3b.toml index 075b1972313..c398af2eefb 100644 --- a/providers/greenpt/models/qwen3.6-35b-a3b.toml +++ b/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 From b11980229871a83f904e91484c8c2995154cb866 Mon Sep 17 00:00:00 2001 From: Cas Burggraaf Date: Tue, 4 Aug 2026 14:58:15 +0200 Subject: [PATCH 7/8] Add deepseek-v4-flash-0731 to GreenPT Generally available on GreenPT at 0.1596 input, 0.399 output and 0.0456 for cached prompt tokens, with the 1M context inherited from the base model. It accepts the full reasoning_effort value set. https://docs.greenpt.ai/model-cards --- .../greenpt/models/deepseek-v4-flash-0731.toml | 14 ++++++++++++++ 1 file changed, 14 insertions(+) create mode 100644 providers/greenpt/models/deepseek-v4-flash-0731.toml diff --git a/providers/greenpt/models/deepseek-v4-flash-0731.toml b/providers/greenpt/models/deepseek-v4-flash-0731.toml new file mode 100644 index 00000000000..aa04728170d --- /dev/null +++ b/providers/greenpt/models/deepseek-v4-flash-0731.toml @@ -0,0 +1,14 @@ +# 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" +reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high"] }] + +[cost] +input = 0.1596 +cache_read = 0.0456 +output = 0.399 From 7c15ffd21f2e8d6984a4638e056b8ef70e04266a Mon Sep 17 00:00:00 2001 From: Cas Burggraaf Date: Tue, 4 Aug 2026 15:05:17 +0200 Subject: [PATCH 8/8] Date deepseek-v4-flash-0731 to its own snapshot The id is the 2026-07-31 snapshot, so inheriting the base model's 2026-04-24 release and update dates would have shown the wrong dates for this endpoint. The remaining inherited fields were checked against production: structured output and tool calling both work, and the 1M context matches the published model card. attachment stays false, since the model card lists no vision capability. --- providers/greenpt/models/deepseek-v4-flash-0731.toml | 2 ++ 1 file changed, 2 insertions(+) diff --git a/providers/greenpt/models/deepseek-v4-flash-0731.toml b/providers/greenpt/models/deepseek-v4-flash-0731.toml index aa04728170d..67886d6c2e5 100644 --- a/providers/greenpt/models/deepseek-v4-flash-0731.toml +++ b/providers/greenpt/models/deepseek-v4-flash-0731.toml @@ -6,6 +6,8 @@ # 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]