From cddcdd671134bab8f35048bde2b0bde07aff8e3f Mon Sep 17 00:00:00 2001
From: Lorenzo Cancellara <lorenzo.cancellara@globalservs.com>
Date: Wed, 5 Aug 2026 15:04:17 +0200
Subject: [PATCH] kimi-for-coding: use indicative costs from base models
 instead of zeroed costs

---
 providers/kimi-for-coding/models/k3-256k.toml          | 10 +++-------
 providers/kimi-for-coding/models/k3.toml               | 10 +++-------
 .../models/kimi-for-coding-highspeed.toml              |  9 +++------
 providers/kimi-for-coding/models/kimi-for-coding.toml  |  9 +++------
 4 files changed, 12 insertions(+), 26 deletions(-)

diff --git a/providers/kimi-for-coding/models/k3-256k.toml b/providers/kimi-for-coding/models/k3-256k.toml
index d28baff592..c4aad1b8af 100644
--- a/providers/kimi-for-coding/models/k3-256k.toml
+++ b/providers/kimi-for-coding/models/k3-256k.toml
@@ -2,7 +2,9 @@
 # endpoint. Unlike full K3 it accepts no video input (image only).
 # reasoning_options per the Kimi Code docs:
 #   reasoning_effort = "low" | "high" | "max" (default "high")
-# Cost is zeroed per this provider's subscription convention.
+# Kimi For Coding is a fixed-fee subscription plan. The cost is inherited
+# from the base model and is the *indicative* pay-per-token equivalent, for
+# usage tracking only, not the actual billing rate.
 base_model = "moonshotai/kimi-k3"
 name = "Kimi K3-256K"
 description = "256K-context version of Kimi K3, reducing token consumption for shorter coding sessions"
@@ -12,12 +14,6 @@ attachment = false
 type = "effort"
 values = ["low", "high", "max"]
 
-[cost]
-input = 0
-output = 0
-cache_read = 0
-cache_write = 0
-
 [limit]
 context = 262_144
 
diff --git a/providers/kimi-for-coding/models/k3.toml b/providers/kimi-for-coding/models/k3.toml
index 3b31cf148d..05075164c4 100644
--- a/providers/kimi-for-coding/models/k3.toml
+++ b/providers/kimi-for-coding/models/k3.toml
@@ -1,7 +1,9 @@
 # reasoning_options mirror the Moonshot AI platform API surface:
 #   thinking.type = "enabled" | "disabled" | "adaptive"
 #   output_config.effort = "low" | "high" | "max"
-# Cost is zeroed per this provider's subscription convention.
+# Kimi For Coding is a fixed-fee subscription plan. The cost is inherited
+# from the base model and is the *indicative* pay-per-token equivalent, for
+# usage tracking only, not the actual billing rate.
 base_model = "moonshotai/kimi-k3"
 attachment = false
 
@@ -11,9 +13,3 @@ type = "toggle"
 [[reasoning_options]]
 type = "effort"
 values = ["low", "high", "max"]
-
-[cost]
-input = 0
-output = 0
-cache_read = 0
-cache_write = 0
diff --git a/providers/kimi-for-coding/models/kimi-for-coding-highspeed.toml b/providers/kimi-for-coding/models/kimi-for-coding-highspeed.toml
index 32d5102804..fb4541f228 100644
--- a/providers/kimi-for-coding/models/kimi-for-coding-highspeed.toml
+++ b/providers/kimi-for-coding/models/kimi-for-coding-highspeed.toml
@@ -1,12 +1,9 @@
+# Kimi For Coding is a fixed-fee subscription plan. The cost is inherited
+# from the base model and is the *indicative* pay-per-token equivalent, for
+# usage tracking only, not the actual billing rate.
 base_model = "moonshotai/kimi-k2.7-code-highspeed"
 name = "Kimi For Coding HighSpeed"
 reasoning_options = []
 
-[cost]
-input = 0
-output = 0
-cache_read = 0
-cache_write = 0
-
 [limit]
 output = 32_768
diff --git a/providers/kimi-for-coding/models/kimi-for-coding.toml b/providers/kimi-for-coding/models/kimi-for-coding.toml
index c6486dd050..96ff773a62 100644
--- a/providers/kimi-for-coding/models/kimi-for-coding.toml
+++ b/providers/kimi-for-coding/models/kimi-for-coding.toml
@@ -1,11 +1,8 @@
+# Kimi For Coding is a fixed-fee subscription plan. The cost is inherited
+# from the base model and is the *indicative* pay-per-token equivalent, for
+# usage tracking only, not the actual billing rate.
 base_model = "moonshotai/kimi-k2.7-code"
 reasoning_options = []
 
-[cost]
-input = 0
-output = 0
-cache_read = 0
-cache_write = 0
-
 [limit]
 output = 32_768