From 5c818fc710a8d5d4a0cabdef5a7c2d98d8ede0fe Mon Sep 17 00:00:00 2001 From: "github-actions[bot]" <41898282+github-actions[bot]@users.noreply.github.com> Date: Mon, 6 Jul 2026 13:09:14 +0000 Subject: [PATCH 1/2] fix: GLM 5.2 on Subconscious, pls? and have bug... --- .../models/subconscious/glm-5.2.toml | 23 +++++++++++++++++++ 1 file changed, 23 insertions(+) create mode 100644 providers/subconscious/models/subconscious/glm-5.2.toml diff --git a/providers/subconscious/models/subconscious/glm-5.2.toml b/providers/subconscious/models/subconscious/glm-5.2.toml new file mode 100644 index 0000000000..619e10761a --- /dev/null +++ b/providers/subconscious/models/subconscious/glm-5.2.toml @@ -0,0 +1,23 @@ +# Subconscious serves GLM-5.2 as subconscious/glm-5.2. +# Pricing (USD per 1M tokens): cached $0.26, input $1.40, output $4.40. +# Reasoning controls: the OpenAI-compatible format toggles thinking via +# chat_template_kwargs.enable_thinking (true|false); the Anthropic Messages +# format uses thinking.type = "enabled"|"disabled" with thinking.budget_tokens +# to cap reasoning. For the OpenAI-compatible endpoint, reasoning streams inline +# within the content field (between think tags). +# Sources (accessed 2026-07-06): +# https://www.subconscious.dev/pricing +# https://docs.subconscious.dev/features/thinking +base_model = "zhipuai/glm-5.2" +interleaved = true + +[[reasoning_options]] +type = "toggle" # API: {"chat_template_kwargs": {"enable_thinking": true | false}} + +[[reasoning_options]] +type = "budget_tokens" # API: {"thinking": {"type": "enabled", "budget_tokens": }} + +[cost] +input = 1.4 +output = 4.4 +cache_read = 0.26 From 645321823c14a2c4806bb771b68e5352b820dc0b Mon Sep 17 00:00:00 2001 From: Aiden Cline Date: Mon, 6 Jul 2026 14:23:16 -0500 Subject: [PATCH 2/2] fix(subconscious): align GLM-5.2 reasoning metadata --- .../subconscious/models/subconscious/glm-5.2.toml | 12 +++++------- 1 file changed, 5 insertions(+), 7 deletions(-) diff --git a/providers/subconscious/models/subconscious/glm-5.2.toml b/providers/subconscious/models/subconscious/glm-5.2.toml index 619e10761a..371e70471d 100644 --- a/providers/subconscious/models/subconscious/glm-5.2.toml +++ b/providers/subconscious/models/subconscious/glm-5.2.toml @@ -1,18 +1,16 @@ # Subconscious serves GLM-5.2 as subconscious/glm-5.2. # Pricing (USD per 1M tokens): cached $0.26, input $1.40, output $4.40. -# Reasoning controls: the OpenAI-compatible format toggles thinking via -# chat_template_kwargs.enable_thinking (true|false); the Anthropic Messages -# format uses thinking.type = "enabled"|"disabled" with thinking.budget_tokens -# to cap reasoning. For the OpenAI-compatible endpoint, reasoning streams inline -# within the content field (between think tags). +# Reasoning controls: Anthropic Messages enables thinking with thinking.type = +# "enabled" and thinking.budget_tokens; omit thinking to disable it. Reasoning +# is returned in separate thinking content blocks. # Sources (accessed 2026-07-06): # https://www.subconscious.dev/pricing +# https://www.subconscious.dev/blog/subconscious-glm-5-2-makes-compact-obsolete # https://docs.subconscious.dev/features/thinking base_model = "zhipuai/glm-5.2" -interleaved = true [[reasoning_options]] -type = "toggle" # API: {"chat_template_kwargs": {"enable_thinking": true | false}} +type = "toggle" # API: omit thinking to disable; {"thinking": {"type": "enabled", "budget_tokens": }} to enable [[reasoning_options]] type = "budget_tokens" # API: {"thinking": {"type": "enabled", "budget_tokens": }}