diff --git a/providers/openrouter/models/deepseek/deepseek-r1-0528.toml b/providers/openrouter/models/deepseek/deepseek-r1-0528.toml
index 1289595d80c..3a47132834b 100644
--- a/providers/openrouter/models/deepseek/deepseek-r1-0528.toml
+++ b/providers/openrouter/models/deepseek/deepseek-r1-0528.toml
@@ -1,26 +1,8 @@
-name = "R1 0528"
+base_model = "deepseek/deepseek-r1-0528"
 description = "DeepSeek reasoning model for multi-step analysis, math, coding, and tools"
-family = "deepseek"
-release_date = "2025-05-28"
-last_updated = "2025-05-28"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = true
-knowledge = "2025-03-31"
-open_weights = true
 reasoning_options = []
 
 [cost]
 input = 0.5
 output = 2.15
 cache_read = 0.35
-
-[limit]
-context = 163_840
-output = 32_768
-
-[modalities]
-input = ["text"]
-output = ["text"]
diff --git a/providers/openrouter/models/deepseek/deepseek-v4-flash-0731.toml b/providers/openrouter/models/deepseek/deepseek-v4-flash-0731.toml
index 9808411a508..8a8d94af0d6 100644
--- a/providers/openrouter/models/deepseek/deepseek-v4-flash-0731.toml
+++ b/providers/openrouter/models/deepseek/deepseek-v4-flash-0731.toml
@@ -10,9 +10,9 @@ type = "effort"
 values = ["low", "high", "max"]
 
 [cost]
-input = 0.0079
+input = 0.0064
 output = 1.28
-cache_read = 0.0079
+cache_read = 0.0064
 
 [limit]
 context = 1_048_576
diff --git a/providers/openrouter/models/deepseek/deepseek-v4-flash.toml b/providers/openrouter/models/deepseek/deepseek-v4-flash.toml
index 7240c8750ef..1e4c4455f50 100644
--- a/providers/openrouter/models/deepseek/deepseek-v4-flash.toml
+++ b/providers/openrouter/models/deepseek/deepseek-v4-flash.toml
@@ -13,9 +13,9 @@ type = "effort"
 values = ["high", "xhigh"]
 
 [cost]
-input = 0.03
+input = 0.0131
 output = 1.28
-cache_read = 0.03
+cache_read = 0.0131
 
 [limit]
 context = 1_048_576
diff --git a/providers/openrouter/models/deepseek/deepseek-v4-pro.toml b/providers/openrouter/models/deepseek/deepseek-v4-pro.toml
index c0bc40d8681..d85e7f57086 100644
--- a/providers/openrouter/models/deepseek/deepseek-v4-pro.toml
+++ b/providers/openrouter/models/deepseek/deepseek-v4-pro.toml
@@ -13,9 +13,9 @@ type = "effort"
 values = ["high", "xhigh"]
 
 [cost]
-input = 0.2871
-output = 0.5742
-cache_read = 0.023925
+input = 0.2088
+output = 0.4176
+cache_read = 0.0174
 
 [limit]
 context = 1_048_576
diff --git a/providers/openrouter/models/meta-llama/llama-3.2-1b-instruct.toml b/providers/openrouter/models/meta-llama/llama-3.2-1b-instruct.toml
index 74ed9344b40..cce4b71c294 100644
--- a/providers/openrouter/models/meta-llama/llama-3.2-1b-instruct.toml
+++ b/providers/openrouter/models/meta-llama/llama-3.2-1b-instruct.toml
@@ -1,15 +1,7 @@
-name = "Llama 3.2 1B Instruct"
+base_model = "meta/llama-3.2-1b-instruct"
 description = "Open Llama instruction model for multilingual chat, reasoning, and coding"
-family = "llama"
-release_date = "2024-09-25"
-last_updated = "2024-09-25"
-attachment = false
-reasoning = false
-temperature = true
 tool_call = false
 structured_output = false
-knowledge = "2023-12-31"
-open_weights = true
 
 [cost]
 input = 0.027
@@ -18,7 +10,3 @@ output = 0.201
 [limit]
 context = 60_000
 output = 54_000
-
-[modalities]
-input = ["text"]
-output = ["text"]
diff --git a/providers/openrouter/models/moonshotai/kimi-k2-0905.toml b/providers/openrouter/models/moonshotai/kimi-k2-0905.toml
index a867b9f2383..adb34cfb895 100644
--- a/providers/openrouter/models/moonshotai/kimi-k2-0905.toml
+++ b/providers/openrouter/models/moonshotai/kimi-k2-0905.toml
@@ -1,24 +1,7 @@
-name = "Kimi K2 0905"
+base_model = "moonshotai/kimi-k2-0905"
 description = "Kimi model for long-context chat, coding, and agentic reasoning"
-family = "kimi-k2"
-release_date = "2025-09-04"
-last_updated = "2025-09-04"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = true
 structured_output = true
-knowledge = "2024-12-31"
-open_weights = true
 
 [cost]
 input = 0.6
 output = 2.5
-
-[limit]
-context = 262_144
-output = 98_304
-
-[modalities]
-input = ["text"]
-output = ["text"]
diff --git a/providers/openrouter/models/moonshotai/kimi-k2-thinking.toml b/providers/openrouter/models/moonshotai/kimi-k2-thinking.toml
index 3ccd5a4f3ef..589eb077162 100644
--- a/providers/openrouter/models/moonshotai/kimi-k2-thinking.toml
+++ b/providers/openrouter/models/moonshotai/kimi-k2-thinking.toml
@@ -8,6 +8,7 @@ field = "reasoning_details"
 [cost]
 input = 0.6
 output = 2.5
+cache_read = 0.15
 
 [limit]
-output = 235_929
+output = 98_304
diff --git a/providers/openrouter/models/moonshotai/kimi-k2.6.toml b/providers/openrouter/models/moonshotai/kimi-k2.6.toml
index 512593e96f8..73e9b9ced62 100644
--- a/providers/openrouter/models/moonshotai/kimi-k2.6.toml
+++ b/providers/openrouter/models/moonshotai/kimi-k2.6.toml
@@ -9,9 +9,9 @@ field = "reasoning_details"
 type = "toggle"
 
 [cost]
-input = 0.4348
+input = 0.4344
 output = 2.45
-cache_read = 0.1202
+cache_read = 0.1077
 
 [limit]
 output = 235_929
diff --git a/providers/openrouter/models/moonshotai/kimi-k3.toml b/providers/openrouter/models/moonshotai/kimi-k3.toml
index 25c97bddc62..9837ed2ea08 100644
--- a/providers/openrouter/models/moonshotai/kimi-k3.toml
+++ b/providers/openrouter/models/moonshotai/kimi-k3.toml
@@ -12,9 +12,9 @@ type = "effort"
 values = ["low", "high", "max"]
 
 [cost]
-input = 0.5
-output = 12
-cache_read = 0.33
+input = 0.9
+output = 14
+cache_read = 0.6
 
 [limit]
 output = 943_718
diff --git a/providers/openrouter/models/qwen/qwen-plus-2025-07-28.toml b/providers/openrouter/models/qwen/qwen-plus-2025-07-28.toml
deleted file mode 100644
index bd96fb94a9c..00000000000
--- a/providers/openrouter/models/qwen/qwen-plus-2025-07-28.toml
+++ /dev/null
@@ -1,29 +0,0 @@
-name = "Qwen Plus 0728"
-description = "Qwen instruction model for multilingual chat, reasoning, and tool use"
-family = "qwen"
-release_date = "2025-09-08"
-last_updated = "2025-09-08"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = true
-structured_output = true
-knowledge = "2025-03-31"
-open_weights = false
-
-[cost]
-input = 0.26
-output = 0.78
-
-[[cost.tiers]]
-tier = { type = "context", size = 256_000 }
-input = 0.78
-output = 2.34
-
-[limit]
-context = 1_000_000
-output = 32_768
-
-[modalities]
-input = ["text"]
-output = ["text"]
diff --git a/providers/openrouter/models/qwen/qwen3-14b.toml b/providers/openrouter/models/qwen/qwen3-14b.toml
index b10030292e3..02217df26ec 100644
--- a/providers/openrouter/models/qwen/qwen3-14b.toml
+++ b/providers/openrouter/models/qwen/qwen3-14b.toml
@@ -21,7 +21,7 @@ input = 0.12
 output = 0.24
 
 [limit]
-context = 131_072
+context = 40_960
 output = 16_384
 
 [modalities]
diff --git a/providers/openrouter/models/qwen/qwen3-235b-a22b-thinking-2507.toml b/providers/openrouter/models/qwen/qwen3-235b-a22b-thinking-2507.toml
index d6dc77cb347..5277b9d5a48 100644
--- a/providers/openrouter/models/qwen/qwen3-235b-a22b-thinking-2507.toml
+++ b/providers/openrouter/models/qwen/qwen3-235b-a22b-thinking-2507.toml
@@ -13,12 +13,12 @@ open_weights = true
 reasoning_options = []
 
 [cost]
-input = 0.23
-output = 2.3
+input = 0.45
+output = 3.5
 
 [limit]
-context = 131_072
-output = 117_964
+context = 128_000
+output = 16_384
 
 [modalities]
 input = ["text"]
diff --git a/providers/openrouter/models/qwen/qwen3-235b-a22b.toml b/providers/openrouter/models/qwen/qwen3-235b-a22b.toml
deleted file mode 100644
index 3f3f59b0214..00000000000
--- a/providers/openrouter/models/qwen/qwen3-235b-a22b.toml
+++ /dev/null
@@ -1,14 +0,0 @@
-# Toggle: reasoning.enabled = true|false
-# https://openrouter.ai/docs/guides/best-practices/reasoning-tokens
-base_model = "alibaba/qwen3-235b-a22b"
-structured_output = false
-
-[[reasoning_options]]
-type = "toggle"
-
-[cost]
-input = 0.455
-output = 1.82
-
-[limit]
-output = 8_192
diff --git a/providers/openrouter/models/qwen/qwen3-30b-a3b-thinking-2507.toml b/providers/openrouter/models/qwen/qwen3-30b-a3b-thinking-2507.toml
deleted file mode 100644
index e89aa371d69..00000000000
--- a/providers/openrouter/models/qwen/qwen3-30b-a3b-thinking-2507.toml
+++ /dev/null
@@ -1,25 +0,0 @@
-name = "Qwen3 30B A3B Thinking 2507"
-description = "Qwen reasoning model for deliberate problem solving, math, and coding"
-family = "qwen"
-release_date = "2025-08-28"
-last_updated = "2025-08-28"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = false
-knowledge = "2025-06-30"
-open_weights = true
-reasoning_options = []
-
-[cost]
-input = 0.2
-output = 2.4
-
-[limit]
-context = 81_920
-output = 32_768
-
-[modalities]
-input = ["text"]
-output = ["text"]
diff --git a/providers/openrouter/models/qwen/qwen3-30b-a3b.toml b/providers/openrouter/models/qwen/qwen3-30b-a3b.toml
index dca225b27ec..d540c4ea0f4 100644
--- a/providers/openrouter/models/qwen/qwen3-30b-a3b.toml
+++ b/providers/openrouter/models/qwen/qwen3-30b-a3b.toml
@@ -10,3 +10,6 @@ type = "toggle"
 [cost]
 input = 0.12
 output = 0.5
+
+[limit]
+context = 40_960
diff --git a/providers/openrouter/models/qwen/qwen3-8b.toml b/providers/openrouter/models/qwen/qwen3-8b.toml
deleted file mode 100644
index 37f8e9c0548..00000000000
--- a/providers/openrouter/models/qwen/qwen3-8b.toml
+++ /dev/null
@@ -1,27 +0,0 @@
-name = "Qwen3 8B"
-description = "Qwen instruction model for multilingual chat, reasoning, and tool use"
-family = "qwen"
-release_date = "2025-04-28"
-last_updated = "2025-04-28"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = false
-knowledge = "2025-03-31"
-open_weights = true
-
-[[reasoning_options]]
-type = "toggle"
-
-[cost]
-input = 0.117
-output = 0.455
-
-[limit]
-context = 131_072
-output = 8_192
-
-[modalities]
-input = ["text"]
-output = ["text"]
diff --git a/providers/openrouter/models/qwen/qwen3-coder-plus.toml b/providers/openrouter/models/qwen/qwen3-coder-plus.toml
deleted file mode 100644
index bdc642f5c82..00000000000
--- a/providers/openrouter/models/qwen/qwen3-coder-plus.toml
+++ /dev/null
@@ -1,25 +0,0 @@
-base_model = "alibaba/qwen3-coder-plus"
-structured_output = true
-
-[cost]
-input = 0.65
-output = 3.25
-cache_read = 0.13
-cache_write = 0.8125
-
-[[cost.tiers]]
-tier = { type = "context", size = 32_000 }
-input = 1.17
-output = 5.85
-cache_read = 0.234
-cache_write = 1.4625
-
-[[cost.tiers]]
-tier = { type = "context", size = 128_000 }
-input = 1.95
-output = 9.75
-cache_read = 0.39
-cache_write = 2.4375
-
-[limit]
-context = 1_000_000
diff --git a/providers/openrouter/models/qwen/qwen3-max-thinking.toml b/providers/openrouter/models/qwen/qwen3-max-thinking.toml
deleted file mode 100644
index 7eb62a6a784..00000000000
--- a/providers/openrouter/models/qwen/qwen3-max-thinking.toml
+++ /dev/null
@@ -1,38 +0,0 @@
-# Toggle: reasoning.enabled = true|false
-# https://openrouter.ai/docs/guides/best-practices/reasoning-tokens
-name = "Qwen3 Max Thinking"
-description = "Qwen reasoning model for deliberate problem solving, math, and coding"
-family = "qwen"
-release_date = "2026-02-09"
-last_updated = "2026-02-09"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = false
-
-[[reasoning_options]]
-type = "toggle"
-
-[cost]
-input = 0.78
-output = 3.9
-
-[[cost.tiers]]
-tier = { type = "context", size = 32_000 }
-input = 1.56
-output = 7.8
-
-[[cost.tiers]]
-tier = { type = "context", size = 128_000 }
-input = 1.95
-output = 9.75
-
-[limit]
-context = 262_144
-output = 65_536
-
-[modalities]
-input = ["text"]
-output = ["text"]
diff --git a/providers/openrouter/models/qwen/qwen3-max.toml b/providers/openrouter/models/qwen/qwen3-max.toml
deleted file mode 100644
index 0a2a33c8ccd..00000000000
--- a/providers/openrouter/models/qwen/qwen3-max.toml
+++ /dev/null
@@ -1,22 +0,0 @@
-base_model = "alibaba/qwen3-max"
-structured_output = true
-
-[cost]
-input = 0.78
-output = 3.9
-cache_read = 0.156
-cache_write = 0.975
-
-[[cost.tiers]]
-tier = { type = "context", size = 32_000 }
-input = 1.56
-output = 7.8
-cache_read = 0.312
-cache_write = 1.95
-
-[[cost.tiers]]
-tier = { type = "context", size = 128_000 }
-input = 1.95
-output = 9.75
-cache_read = 0.39
-cache_write = 2.4375
diff --git a/providers/openrouter/models/qwen/qwen3-next-80b-a3b-thinking.toml b/providers/openrouter/models/qwen/qwen3-next-80b-a3b-thinking.toml
index 14e222d894d..35f13bee686 100644
--- a/providers/openrouter/models/qwen/qwen3-next-80b-a3b-thinking.toml
+++ b/providers/openrouter/models/qwen/qwen3-next-80b-a3b-thinking.toml
@@ -8,3 +8,4 @@ output = 1.2
 
 [limit]
 context = 262_144
+output = 235_929
diff --git a/providers/openrouter/models/qwen/qwen3-vl-235b-a22b-thinking.toml b/providers/openrouter/models/qwen/qwen3-vl-235b-a22b-thinking.toml
deleted file mode 100644
index b98df3d1d6c..00000000000
--- a/providers/openrouter/models/qwen/qwen3-vl-235b-a22b-thinking.toml
+++ /dev/null
@@ -1,7 +0,0 @@
-base_model = "alibaba/qwen3-vl-235b-a22b-thinking"
-description = "Qwen vision-language model for visual reasoning, documents, and agent tasks"
-reasoning_options = []
-
-[cost]
-input = 0.4
-output = 4
diff --git a/providers/openrouter/models/qwen/qwen3-vl-30b-a3b-instruct.toml b/providers/openrouter/models/qwen/qwen3-vl-30b-a3b-instruct.toml
index 3f78c99ff08..2117029a6e4 100644
--- a/providers/openrouter/models/qwen/qwen3-vl-30b-a3b-instruct.toml
+++ b/providers/openrouter/models/qwen/qwen3-vl-30b-a3b-instruct.toml
@@ -1,24 +1,13 @@
-name = "Qwen3 VL 30B A3B Instruct"
+base_model = "alibaba/qwen3-vl-30b-a3b-instruct"
 description = "Qwen vision-language model for visual reasoning, documents, and agent tasks"
-family = "qwen"
-release_date = "2025-10-06"
-last_updated = "2025-10-06"
-attachment = true
-reasoning = false
-temperature = true
-tool_call = true
 structured_output = true
-knowledge = "2025-03-31"
-open_weights = true
 
 [cost]
 input = 0.15
 output = 0.6
 
 [limit]
-context = 262_144
 output = 16_384
 
 [modalities]
 input = ["text", "image"]
-output = ["text"]
diff --git a/providers/openrouter/models/qwen/qwen3-vl-30b-a3b-thinking.toml b/providers/openrouter/models/qwen/qwen3-vl-30b-a3b-thinking.toml
index cf36c7ec114..e55d4bea8da 100644
--- a/providers/openrouter/models/qwen/qwen3-vl-30b-a3b-thinking.toml
+++ b/providers/openrouter/models/qwen/qwen3-vl-30b-a3b-thinking.toml
@@ -13,12 +13,12 @@ open_weights = true
 reasoning_options = []
 
 [cost]
-input = 0.2
-output = 2.4
+input = 0.29
+output = 1
 
 [limit]
 context = 262_144
-output = 32_768
+output = 235_929
 
 [modalities]
 input = ["text", "image"]
diff --git a/providers/openrouter/models/qwen/qwen3-vl-32b-instruct.toml b/providers/openrouter/models/qwen/qwen3-vl-32b-instruct.toml
deleted file mode 100644
index 55f43925d87..00000000000
--- a/providers/openrouter/models/qwen/qwen3-vl-32b-instruct.toml
+++ /dev/null
@@ -1,23 +0,0 @@
-name = "Qwen3 VL 32B Instruct"
-description = "Qwen vision-language model for visual reasoning, documents, and agent tasks"
-family = "qwen"
-release_date = "2025-10-23"
-last_updated = "2025-10-23"
-attachment = true
-reasoning = false
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.104
-output = 0.416
-
-[limit]
-context = 131_072
-output = 32_768
-
-[modalities]
-input = ["text", "image"]
-output = ["text"]
diff --git a/providers/openrouter/models/qwen/qwen3-vl-8b-instruct.toml b/providers/openrouter/models/qwen/qwen3-vl-8b-instruct.toml
index 583e8c86c58..7083577667c 100644
--- a/providers/openrouter/models/qwen/qwen3-vl-8b-instruct.toml
+++ b/providers/openrouter/models/qwen/qwen3-vl-8b-instruct.toml
@@ -11,12 +11,13 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.117
-output = 0.455
+input = 0.25
+output = 0.75
+cache_read = 0.12
 
 [limit]
 context = 262_144
-output = 32_768
+output = 235_929
 
 [modalities]
 input = ["image", "text"]
diff --git a/providers/openrouter/models/qwen/qwen3-vl-8b-thinking.toml b/providers/openrouter/models/qwen/qwen3-vl-8b-thinking.toml
deleted file mode 100644
index c429a59879e..00000000000
--- a/providers/openrouter/models/qwen/qwen3-vl-8b-thinking.toml
+++ /dev/null
@@ -1,24 +0,0 @@
-name = "Qwen3 VL 8B Thinking"
-description = "Qwen vision-language model for visual reasoning, documents, and agent tasks"
-family = "qwen"
-release_date = "2025-10-14"
-last_updated = "2025-10-14"
-attachment = true
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-reasoning_options = []
-
-[cost]
-input = 0.18
-output = 2.1
-
-[limit]
-context = 131_072
-output = 32_768
-
-[modalities]
-input = ["image", "text"]
-output = ["text"]
diff --git a/providers/openrouter/models/qwen/qwen3.6-max-preview.toml b/providers/openrouter/models/qwen/qwen3.6-max-preview.toml
deleted file mode 100644
index 635c24238b4..00000000000
--- a/providers/openrouter/models/qwen/qwen3.6-max-preview.toml
+++ /dev/null
@@ -1,18 +0,0 @@
-# Toggle: reasoning.enabled = true|false
-# https://openrouter.ai/docs/guides/best-practices/reasoning-tokens
-base_model = "alibaba/qwen3.6-max-preview"
-structured_output = true
-
-[[reasoning_options]]
-type = "toggle"
-
-[cost]
-input = 1.027
-output = 6.162
-cache_write = 1.28375
-
-[[cost.tiers]]
-tier = { type = "context", size = 128_000 }
-input = 1.58
-output = 9.48
-cache_write = 1.975
diff --git a/providers/openrouter/models/tencent/hy3.toml b/providers/openrouter/models/tencent/hy3.toml
index 80ecfdd3aa8..f61fa3175e9 100644
--- a/providers/openrouter/models/tencent/hy3.toml
+++ b/providers/openrouter/models/tencent/hy3.toml
@@ -6,9 +6,9 @@ type = "effort"
 values = ["none", "low", "high"]
 
 [cost]
-input = 0.132
-output = 0.528
-cache_read = 0.033
+input = 0.0825
+output = 0.33
+cache_read = 0.020625
 
 [limit]
 context = 262_144
diff --git a/providers/openrouter/models/tencent/hy4-preview.toml b/providers/openrouter/models/tencent/hy4-preview.toml
index e39f8986b6d..4ceb054717c 100644
--- a/providers/openrouter/models/tencent/hy4-preview.toml
+++ b/providers/openrouter/models/tencent/hy4-preview.toml
@@ -7,9 +7,9 @@ type = "effort"
 values = ["none", "low", "high"]
 
 [cost]
-input = 0.834
-output = 2.501
-cache_read = 0.042
+input = 0.7506
+output = 2.2509
+cache_read = 0.0378
 
 [limit]
 context = 1_048_576
diff --git a/providers/openrouter/models/z-ai/glm-5.2.toml b/providers/openrouter/models/z-ai/glm-5.2.toml
index 3c763f23759..a939baf95a4 100644
--- a/providers/openrouter/models/z-ai/glm-5.2.toml
+++ b/providers/openrouter/models/z-ai/glm-5.2.toml
@@ -14,7 +14,7 @@ values = ["high", "xhigh"]
 
 [cost]
 input = 0.06
-output = 4.2
+output = 8
 cache_read = 0.059
 
 [limit]
diff --git a/providers/openrouter/models/z-ai/glm-5.3.toml b/providers/openrouter/models/z-ai/glm-5.3.toml
index 9c8d6c62097..9e47ddd7ade 100644
--- a/providers/openrouter/models/z-ai/glm-5.3.toml
+++ b/providers/openrouter/models/z-ai/glm-5.3.toml
@@ -7,7 +7,7 @@ values = ["low", "high", "max"]
 
 [cost]
 input = 0.04
-output = 7
+output = 4.8
 cache_read = 0.039
 
 [limit]
diff --git a/providers/openrouter/models/~deepseek/deepseek-flash-latest.toml b/providers/openrouter/models/~deepseek/deepseek-flash-latest.toml
index f667aafb6b3..8467bf2a795 100644
--- a/providers/openrouter/models/~deepseek/deepseek-flash-latest.toml
+++ b/providers/openrouter/models/~deepseek/deepseek-flash-latest.toml
@@ -21,8 +21,8 @@ values = ["low", "high", "max"]
 
 [cost]
 input = 0.016
-output = 1.2
-cache_read = 0.01
+output = 0.6
+cache_read = 0.005
 
 [limit]
 context = 1_048_576
diff --git a/providers/openrouter/models/~deepseek/deepseek-v4-flash-latest.toml b/providers/openrouter/models/~deepseek/deepseek-v4-flash-latest.toml
index 4b1d1581a4c..33aff61ecd8 100644
--- a/providers/openrouter/models/~deepseek/deepseek-v4-flash-latest.toml
+++ b/providers/openrouter/models/~deepseek/deepseek-v4-flash-latest.toml
@@ -20,9 +20,9 @@ type = "effort"
 values = ["low", "high", "max"]
 
 [cost]
-input = 0.0079
-output = 1.28
-cache_read = 0.0079
+input = 0.00512
+output = 0.01467
+cache_read = 0.00512
 
 [limit]
 context = 1_048_576
diff --git a/providers/openrouter/models/~moonshotai/kimi-latest.toml b/providers/openrouter/models/~moonshotai/kimi-latest.toml
index a60443e3e88..c80b74d4765 100644
--- a/providers/openrouter/models/~moonshotai/kimi-latest.toml
+++ b/providers/openrouter/models/~moonshotai/kimi-latest.toml
@@ -20,9 +20,9 @@ type = "effort"
 values = ["low", "high", "max"]
 
 [cost]
-input = 0.49
-output = 13
-cache_read = 0.3
+input = 0.45
+output = 14
+cache_read = 0.29
 
 [limit]
 context = 1_048_576
diff --git a/providers/openrouter/models/~z-ai/glm-flash-latest.toml b/providers/openrouter/models/~z-ai/glm-flash-latest.toml
index 231a4cb889e..828755c4b6d 100644
--- a/providers/openrouter/models/~z-ai/glm-flash-latest.toml
+++ b/providers/openrouter/models/~z-ai/glm-flash-latest.toml
@@ -16,7 +16,7 @@ values = ["low", "high", "max"]
 
 [cost]
 input = 0.032
-output = 0.264718
+output = 0.140682
 cache_read = 0.02
 
 [limit]