From 29934c238e38159130b66402c092c8fcb6f37799 Mon Sep 17 00:00:00 2001 From: Mikael Hugo Date: Thu, 20 Aug 2026 04:13:47 +0000 Subject: [PATCH 1/2] fix(ollama-cloud): correct context_length for 3 models Values taken from the provider itself via ollama.com/api/show, which reports model_info..context_length. glm-5.2 976_000 -> 1_000_000 minimax-m3 512_000 -> 524_288 nemotron-3-nano:30b 1048576 -> 262144 nemotron-3-nano:30b is the consequential one: the recorded value is 4x the real window, so a client sizing requests from it overruns and the provider rejects them. --- providers/ollama-cloud/models/glm-5.2.toml | 2 +- providers/ollama-cloud/models/minimax-m3.toml | 2 +- providers/ollama-cloud/models/nemotron-3-nano:30b.toml | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) diff --git a/providers/ollama-cloud/models/glm-5.2.toml b/providers/ollama-cloud/models/glm-5.2.toml index dcb62d38c8a..ba50332e990 100644 --- a/providers/ollama-cloud/models/glm-5.2.toml +++ b/providers/ollama-cloud/models/glm-5.2.toml @@ -9,4 +9,4 @@ values = ["high", "max"] field = "reasoning_content" [limit] -context = 976_000 \ No newline at end of file +context = 1_000_000 \ No newline at end of file diff --git a/providers/ollama-cloud/models/minimax-m3.toml b/providers/ollama-cloud/models/minimax-m3.toml index c8155d651db..4fd0915dbee 100644 --- a/providers/ollama-cloud/models/minimax-m3.toml +++ b/providers/ollama-cloud/models/minimax-m3.toml @@ -13,7 +13,7 @@ open_weights = true [limit] -context = 512_000 +context = 524_288 output = 131_072 [modalities] diff --git a/providers/ollama-cloud/models/nemotron-3-nano:30b.toml b/providers/ollama-cloud/models/nemotron-3-nano:30b.toml index 179fd8e2beb..48b4a68a4c1 100644 --- a/providers/ollama-cloud/models/nemotron-3-nano:30b.toml +++ b/providers/ollama-cloud/models/nemotron-3-nano:30b.toml @@ -10,7 +10,7 @@ last_updated = "2026-01-19" open_weights = true [limit] -context = 1048576 +context = 262144 output = 131072 [modalities] From 99ea58df9af3a57c9ca5908711544cdf77ee67e3 Mon Sep 17 00:00:00 2001 From: Mikael Hugo Date: Thu, 20 Aug 2026 04:28:44 +0000 Subject: [PATCH 2/2] feat(ollama-cloud): add three tags served but not yet catalogued GET https://ollama.com/v1/models lists these three; the database has no entry. deepseek-v4-pro:0813 context 1048576 deepseek-v4-pro:preview context 524288 <- half the stable tag deepseek-v4-flash:preview context 1048576 Contexts read from https://ollama.com/api/show, the same source the existing deepseek-v4-flash:0731 entry cites. Output mirrors context, following every other deepseek entry under this provider; api/show reports no max output, so that field is convention rather than API-verified. --- .../models/deepseek-v4-flash:preview.toml | 18 ++++++++++++++++++ .../models/deepseek-v4-pro:0813.toml | 18 ++++++++++++++++++ .../models/deepseek-v4-pro:preview.toml | 19 +++++++++++++++++++ 3 files changed, 55 insertions(+) create mode 100644 providers/ollama-cloud/models/deepseek-v4-flash:preview.toml create mode 100644 providers/ollama-cloud/models/deepseek-v4-pro:0813.toml create mode 100644 providers/ollama-cloud/models/deepseek-v4-pro:preview.toml diff --git a/providers/ollama-cloud/models/deepseek-v4-flash:preview.toml b/providers/ollama-cloud/models/deepseek-v4-flash:preview.toml new file mode 100644 index 00000000000..ae5a693ccad --- /dev/null +++ b/providers/ollama-cloud/models/deepseek-v4-flash:preview.toml @@ -0,0 +1,18 @@ +# Preview tag of deepseek-v4-flash, served by Ollama Cloud and listed by GET /v1/models. +# API-verified context: https://ollama.com/api/show (deepseek4.context_length = 1048576) +name = "deepseek-v4-flash:preview" +description = "Preview release of the fast DeepSeek model for efficient chat, coding help, and agent loops" +family = "deepseek-flash" +attachment = false +reasoning = true +reasoning_options = [{ type = "toggle" }, { type = "effort", values = ["high", "max"] }] +tool_call = true +open_weights = true + +[limit] +context = 1048576 +output = 1048576 + +[modalities] +input = ["text"] +output = ["text"] diff --git a/providers/ollama-cloud/models/deepseek-v4-pro:0813.toml b/providers/ollama-cloud/models/deepseek-v4-pro:0813.toml new file mode 100644 index 00000000000..040dcdc009d --- /dev/null +++ b/providers/ollama-cloud/models/deepseek-v4-pro:0813.toml @@ -0,0 +1,18 @@ +# Dated tag of deepseek-v4-pro, served by Ollama Cloud and listed by GET /v1/models. +# API-verified context: https://ollama.com/api/show (deepseek4.context_length = 1048576) +name = "deepseek-v4-pro:0813" +description = "Dated release of the flagship DeepSeek model for coding, reasoning, and agentic work" +family = "deepseek-thinking" +attachment = false +reasoning = true +reasoning_options = [{ type = "toggle" }, { type = "effort", values = ["high", "max"] }] +tool_call = true +open_weights = true + +[limit] +context = 1048576 +output = 1048576 + +[modalities] +input = ["text"] +output = ["text"] diff --git a/providers/ollama-cloud/models/deepseek-v4-pro:preview.toml b/providers/ollama-cloud/models/deepseek-v4-pro:preview.toml new file mode 100644 index 00000000000..13c8846138c --- /dev/null +++ b/providers/ollama-cloud/models/deepseek-v4-pro:preview.toml @@ -0,0 +1,19 @@ +# Preview tag of deepseek-v4-pro, served by Ollama Cloud and listed by GET /v1/models. +# NOTE: this tag has HALF the context of the stable tag. +# API-verified context: https://ollama.com/api/show (deepseek4.context_length = 524288) +name = "deepseek-v4-pro:preview" +description = "Preview release of the flagship DeepSeek model for coding, reasoning, and agentic work" +family = "deepseek-thinking" +attachment = false +reasoning = true +reasoning_options = [{ type = "toggle" }, { type = "effort", values = ["high", "max"] }] +tool_call = true +open_weights = true + +[limit] +context = 524288 +output = 524288 + +[modalities] +input = ["text"] +output = ["text"]