feat: add RainDragon provider with Gemma 4 26B and 31B mappings

mappings/raindragon/gemma-4-26b.yaml

@@ -0,0 +1,16 @@
+model_id: gemma-4-26b
+provider_id: raindragon
+provider_model_name: google/gemma-4-26B-A4B-it
+pricing:
+  input_per_million: 0.075
+  output_per_million: 0.35
+  cache_read_per_million: 0.03
+  currency: USD
+rate_limits:
+  requests_per_minute: null
+  tokens_per_minute: null
+context_window_override: 204800
+available_regions:
+  - us-east-1
+created_at: "2026-10-07T00:00:00.000Z"
+updated_at: "2026-10-07T00:00:00.000Z"

mappings/raindragon/gemma-4-31b.yaml

@@ -0,0 +1,16 @@
+model_id: gemma-4-31b
+provider_id: raindragon
+provider_model_name: google/gemma-4-31B-it
+pricing:
+  input_per_million: 0.13
+  output_per_million: 0.38
+  cache_read_per_million: 0.03
+  currency: USD
+rate_limits:
+  requests_per_minute: null
+  tokens_per_minute: null
+context_window_override: 65536
+available_regions:
+  - us-east-1
+created_at: "2026-10-07T00:00:00.000Z"
+updated_at: "2026-10-07T00:00:00.000Z"

providers/raindragon.yaml

@@ -0,0 +1,10 @@
+id: raindragon
+name: RainDragon
+description: US-hosted inference for open-weight models with an OpenAI-compatible API, pay-per-token pricing, discounted cached input, and no prompt or response storage.
+api_base_url: https://inference.raindragon.ai/v1
+auth_type: bearer
+regions:
+  - us-east-1
+compatibility: openai
+created_at: "2026-10-07T00:00:00.000Z"
+updated_at: "2026-10-07T00:00:00.000Z"