fix(llm): capture Venice top-level costs

lib/components/fabro-llm/src/codec/openai_compatible/response.rs

@@ -63,7 +63,9 @@ pub(super) fn decode_response(
 
     let wire_usage = api_resp.usage.as_ref();
     let usage = wire_usage.map_or_else(TokenCounts::default, ApiUsage::token_counts);
-    let cost_usd = wire_usage.and_then(|u| u.cost);
+    let cost_usd = wire_usage
+        .and_then(|usage| usage.cost)
+        .or_else(|| api_resp.cost.as_ref().and_then(|cost| cost.usd));
     let cost_source = translate::authoritative_cost_source(cost_usd);
 
     Ok(Response {

lib/components/fabro-llm/src/codec/openai_compatible/stream.rs

@@ -29,8 +29,8 @@ pub(super) struct StreamState {
     /// True after `finish_events()` has run (guards against duplicates).
     finished:              bool,
     rate_limit:            Option<RateLimitInfo>,
-    /// In-band USD cost from the usage chunk (OpenRouter), surfaced as
-    /// authoritative on the final response.
+    /// In-band USD cost from the response, surfaced as authoritative on the
+    /// final response.
     cost_usd:              Option<f64>,
 }
 
@@ -72,9 +72,13 @@ impl StreamState {
         // Capture usage if present (often in a dedicated chunk).
         if let Some(usage) = &chunk.usage {
             self.usage = usage.token_counts();
-            // Keep a previously seen cost when a later usage chunk omits it.
-            self.cost_usd = usage.cost.or(self.cost_usd);
         }
+        let cost_usd = chunk
+            .usage
+            .as_ref()
+            .and_then(|usage| usage.cost)
+            .or_else(|| chunk.cost.as_ref().and_then(|cost| cost.usd));
+        self.cost_usd = cost_usd.or(self.cost_usd);
 
         let choices = chunk.choices.as_mut()?;
         let choice = choices.first_mut()?;

lib/components/fabro-llm/src/codec/openai_compatible/wire.rs

@@ -125,6 +125,12 @@ pub(super) struct ApiResponse {
     pub model:   String,
     pub choices: Vec<ApiChoice>,
     pub usage:   Option<ApiUsage>,
+    pub cost:    Option<ApiCost>,
+}
+
+#[derive(serde::Deserialize)]
+pub(super) struct ApiCost {
+    pub usd: Option<f64>,
 }
 
 #[derive(serde::Deserialize)]
@@ -377,6 +383,7 @@ pub(super) struct StreamChunk {
     pub model:   Option<String>,
     pub choices: Option<Vec<StreamChoice>>,
     pub usage:   Option<ApiUsage>,
+    pub cost:    Option<ApiCost>,
 }
 
 #[derive(serde::Deserialize)]

lib/components/fabro-llm/tests/it/wire/openai_compatible.rs

@@ -795,6 +795,31 @@ async fn decode_usage_openrouter_cost_and_cache_write() {
     fabro_test::fabro_json_snapshot!(response);
 }
 
+/// Venice reports authoritative USD cost in a top-level object rather than
+/// the OpenRouter `usage.cost` field.
+#[tokio::test]
+async fn decode_usage_venice_top_level_cost() {
+    let response = decode_response(serde_json::json!({
+        "id": "chatcmpl_venice_test",
+        "object": "chat.completion",
+        "created": CREATED_TS,
+        "model": MODEL,
+        "choices": [{
+            "index": 0,
+            "message": {"role": "assistant", "content": "ok"},
+            "finish_reason": "stop"
+        }],
+        "cost": {"usd": 0.00042, "diem": 0.0},
+        "usage": {
+            "prompt_tokens": 12,
+            "completion_tokens": 2,
+            "total_tokens": 14
+        }
+    }))
+    .await;
+    fabro_test::fabro_json_snapshot!(response);
+}
+
 // ---------------------------------------------------------------------------
 // Stream
 // ---------------------------------------------------------------------------
@@ -841,6 +866,20 @@ async fn stream_usage_openrouter_cost() {
     fabro_test::fabro_json_snapshot!(events);
 }
 
+/// Venice streams authoritative USD cost in a top-level object on the usage
+/// chunk.
+#[tokio::test]
+async fn stream_usage_venice_top_level_cost() {
+    let sse = support::sse_data_transcript(&[
+        r#"{"id":"chatcmpl_venice_stream","object":"chat.completion.chunk","created":1700000000,"model":"test-model","choices":[{"index":0,"delta":{"role":"assistant","content":"Hi"},"finish_reason":null}]}"#,
+        r#"{"id":"chatcmpl_venice_stream","object":"chat.completion.chunk","created":1700000000,"model":"test-model","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}"#,
+        r#"{"id":"chatcmpl_venice_stream","object":"chat.completion.chunk","created":1700000000,"model":"test-model","choices":[],"cost":{"usd":0.00031,"diem":0.0},"usage":{"prompt_tokens":12,"completion_tokens":2,"total_tokens":14}}"#,
+        "[DONE]",
+    ]);
+    let (_capture, events) = stream_capture(&base_request(MODEL), &sse).await;
+    fabro_test::fabro_json_snapshot!(events);
+}
+
 #[tokio::test]
 async fn stream_tool_call_deltas() {
     let sse = support::sse_data_transcript(&[

lib/components/fabro-llm/tests/it/wire/snapshots/it__wire__openai_compatible__decode_usage_venice_top_level_cost.snap

@@ -0,0 +1,55 @@
+---
+source: lib/components/fabro-llm/tests/it/wire/openai_compatible.rs
+expression: rendered
+---
+{
+  "id": "chatcmpl_venice_test",
+  "model": "test-model",
+  "provider": "openai-compatible",
+  "message": {
+    "role": "assistant",
+    "content": [
+      {
+        "kind": "text",
+        "data": "ok"
+      }
+    ]
+  },
+  "finish_reason": "stop",
+  "usage": {
+    "input_tokens": 12,
+    "output_tokens": 2,
+    "reasoning_tokens": 0,
+    "cache_read_tokens": 0,
+    "cache_write_tokens": 0
+  },
+  "raw": {
+    "id": "chatcmpl_venice_test",
+    "object": "chat.completion",
+    "created": 1700000000,
+    "model": "test-model",
+    "choices": [
+      {
+        "index": 0,
+        "message": {
+          "role": "assistant",
+          "content": "ok"
+        },
+        "finish_reason": "stop"
+      }
+    ],
+    "cost": {
+      "usd": 0.00042,
+      "diem": 0.0
+    },
+    "usage": {
+      "prompt_tokens": 12,
+      "completion_tokens": 2,
+      "total_tokens": 14
+    }
+  },
+  "warnings": [],
+  "rate_limit": null,
+  "cost_usd": 0.00042,
+  "cost_source": "authoritative"
+}

lib/components/fabro-llm/tests/it/wire/snapshots/it__wire__openai_compatible__stream_usage_venice_top_level_cost.snap

@@ -0,0 +1,60 @@
+---
+source: lib/components/fabro-llm/tests/it/wire/openai_compatible.rs
+expression: rendered
+---
+[
+  {
+    "type": "stream_start"
+  },
+  {
+    "type": "text_start",
+    "text_id": null
+  },
+  {
+    "type": "text_delta",
+    "delta": "Hi",
+    "text_id": null
+  },
+  {
+    "type": "text_end",
+    "text_id": null
+  },
+  {
+    "type": "finish",
+    "finish_reason": "stop",
+    "usage": {
+      "input_tokens": 12,
+      "output_tokens": 2,
+      "reasoning_tokens": 0,
+      "cache_read_tokens": 0,
+      "cache_write_tokens": 0
+    },
+    "response": {
+      "id": "chatcmpl_venice_stream",
+      "model": "test-model",
+      "provider": "openai-compatible",
+      "message": {
+        "role": "assistant",
+        "content": [
+          {
+            "kind": "text",
+            "data": "Hi"
+          }
+        ]
+      },
+      "finish_reason": "stop",
+      "usage": {
+        "input_tokens": 12,
+        "output_tokens": 2,
+        "reasoning_tokens": 0,
+        "cache_read_tokens": 0,
+        "cache_write_tokens": 0
+      },
+      "raw": null,
+      "warnings": [],
+      "rate_limit": null,
+      "cost_usd": 0.00031,
+      "cost_source": "authoritative"
+    }
+  }
+]