@opencode-ai/ai 0.0.0-next-16184 → 0.0.0-next-16187

package/dist/protocols/anthropic-messages.d.ts

@@ -395,18 +395,36 @@ export declare const protocol: Protocol<{
     } | undefined;
     readonly message?: {
         readonly usage?: {
+            readonly [x: string]: unknown;
+            readonly server_tool_use?: {
+                readonly [x: string]: unknown;
+                readonly web_search_requests?: number | undefined;
+            } | null | undefined;
             readonly input_tokens?: number | undefined;
             readonly output_tokens?: number | undefined;
             readonly cache_creation_input_tokens?: number | null | undefined;
             readonly cache_read_input_tokens?: number | null | undefined;
+            readonly output_tokens_details?: {
+                readonly [x: string]: unknown;
+                readonly thinking_tokens?: number | undefined;
+            } | null | undefined;
         } | undefined;
     } | undefined;
     readonly index?: number | undefined;
     readonly usage?: {
+        readonly [x: string]: unknown;
+        readonly server_tool_use?: {
+            readonly [x: string]: unknown;
+            readonly web_search_requests?: number | undefined;
+        } | null | undefined;
         readonly input_tokens?: number | undefined;
         readonly output_tokens?: number | undefined;
         readonly cache_creation_input_tokens?: number | null | undefined;
         readonly cache_read_input_tokens?: number | null | undefined;
+        readonly output_tokens_details?: {
+            readonly [x: string]: unknown;
+            readonly thinking_tokens?: number | undefined;
+        } | null | undefined;
     } | undefined;
     readonly content_block?: {
         readonly type: string;

package/dist/protocols/anthropic-messages.js

@@ -4,7 +4,7 @@ import { Auth } from "../route/auth";
 import { Endpoint } from "../route/endpoint";
 import { Framing } from "../route/framing";
 import { Protocol } from "../route/protocol";
-import { LLMError, LLMEvent, Usage, } from "../schema";
+import { LLMError, LLMEvent, mergeJsonRecords, Usage, } from "../schema";
 import { JsonObject, optionalArray, optionalNull, ProviderShared } from "./shared";
 import { classifyProviderFailure } from "../provider-error";
 import * as Cache from "./utils/cache";
@@ -161,12 +161,14 @@ const AnthropicBodyFields = {
     output_config: Schema.optional(AnthropicOutputConfig),
 };
 export const AnthropicMessagesBody = Schema.Struct(AnthropicBodyFields);
-const AnthropicUsage = Schema.Struct({
+const AnthropicUsage = Schema.StructWithRest(Schema.Struct({
     input_tokens: Schema.optional(Schema.Number),
     output_tokens: Schema.optional(Schema.Number),
     cache_creation_input_tokens: optionalNull(Schema.Number),
     cache_read_input_tokens: optionalNull(Schema.Number),
-});
+    server_tool_use: optionalNull(Schema.StructWithRest(Schema.Struct({ web_search_requests: Schema.optional(Schema.Number) }), [Schema.Record(Schema.String, Schema.Unknown)])),
+    output_tokens_details: optionalNull(Schema.StructWithRest(Schema.Struct({ thinking_tokens: Schema.optional(Schema.Number) }), [Schema.Record(Schema.String, Schema.Unknown)])),
+}), [Schema.Record(Schema.String, Schema.Unknown)]);
 const AnthropicStreamBlock = Schema.Struct({
     type: Schema.String,
     id: Schema.optional(Schema.String),
@@ -535,9 +537,8 @@ const mapFinishReason = (reason) => {
 // `input_tokens` is the *non-cached* count per the Messages API docs, with
 // cache reads and writes as separate fields. We sum them to derive the
 // inclusive `inputTokens` the rest of the contract expects. Extended
-// thinking tokens are *not* broken out by Anthropic — they're billed as
-// part of `output_tokens`, so `reasoningTokens` stays `undefined` and
-// `outputTokens` carries the combined total.
+// thinking tokens are included in `output_tokens`; newer responses also
+// expose that subset through `output_tokens_details.thinking_tokens`.
 const mapUsage = (usage) => {
     if (!usage)
         return undefined;
@@ -551,6 +552,7 @@ const mapUsage = (usage) => {
         nonCachedInputTokens: nonCached,
         cacheReadInputTokens: cacheRead,
         cacheWriteInputTokens: cacheWrite,
+        reasoningTokens: usage.output_tokens_details?.thinking_tokens,
         totalTokens: ProviderShared.totalTokens(inputTokens, usage.output_tokens, undefined),
         providerMetadata: { anthropic: usage },
     });
@@ -570,18 +572,17 @@ const mergeUsage = (left, right) => {
     const cacheWriteInputTokens = right.cacheWriteInputTokens ?? left.cacheWriteInputTokens;
     const inputTokens = ProviderShared.sumTokens(nonCachedInputTokens, cacheReadInputTokens, cacheWriteInputTokens);
     const outputTokens = right.outputTokens ?? left.outputTokens;
+    const reasoningTokens = right.reasoningTokens ?? left.reasoningTokens;
     return new Usage({
         inputTokens,
         outputTokens,
         nonCachedInputTokens,
         cacheReadInputTokens,
         cacheWriteInputTokens,
+        reasoningTokens,
         totalTokens: ProviderShared.totalTokens(inputTokens, outputTokens, undefined),
         providerMetadata: {
-            anthropic: {
-                ...left.providerMetadata?.["anthropic"],
-                ...right.providerMetadata?.["anthropic"],
-            },
+            anthropic: mergeJsonRecords(left.providerMetadata?.["anthropic"], right.providerMetadata?.["anthropic"]) ?? {},
         },
     });
 };

package/dist/protocols/open-responses.d.ts

@@ -495,14 +495,14 @@ export declare const protocol: Protocol<{
         readonly usage?: {
             readonly input_tokens?: number | undefined;
             readonly output_tokens?: number | undefined;
+            readonly output_tokens_details?: {
+                readonly reasoning_tokens?: number | undefined;
+            } | null | undefined;
             readonly total_tokens?: number | undefined;
             readonly input_tokens_details?: {
                 readonly cached_tokens?: number | undefined;
                 readonly cache_write_tokens?: number | undefined;
             } | null | undefined;
-            readonly output_tokens_details?: {
-                readonly reasoning_tokens?: number | undefined;
-            } | null | undefined;
         } | null | undefined;
         readonly service_tier?: string | null | undefined;
         readonly incomplete_details?: {

package/dist/protocols/openai-responses.d.ts

@@ -220,14 +220,14 @@ export declare const protocol: Protocol<{
         readonly usage?: {
             readonly input_tokens?: number | undefined;
             readonly output_tokens?: number | undefined;
+            readonly output_tokens_details?: {
+                readonly reasoning_tokens?: number | undefined;
+            } | null | undefined;
             readonly total_tokens?: number | undefined;
             readonly input_tokens_details?: {
                 readonly cached_tokens?: number | undefined;
                 readonly cache_write_tokens?: number | undefined;
             } | null | undefined;
-            readonly output_tokens_details?: {
-                readonly reasoning_tokens?: number | undefined;
-            } | null | undefined;
         } | null | undefined;
         readonly service_tier?: string | null | undefined;
         readonly incomplete_details?: {

package/dist/schema/events.d.ts

@@ -47,9 +47,9 @@ declare const Usage_base: Schema.Class<Usage, Schema.Struct<{
  * - Anthropic and Bedrock report the input breakdown natively: Anthropic's
  *   `input_tokens` and Bedrock's `inputTokens` are non-cached only. Their
  *   mappers sum the breakdown to derive the inclusive `inputTokens`.
- *   Anthropic does *not* break extended-thinking out of `output_tokens`, so
- *   `reasoningTokens` is `undefined` and `outputTokens` carries the
- *   combined total — a documented limitation of the Anthropic API.
+ *   Anthropic's `outputTokens` includes extended thinking. Newer responses
+ *   expose that subset as `output_tokens_details.thinking_tokens`, which maps
+ *   to `reasoningTokens`; older responses leave it undefined.
  *
  * `providerMetadata` always carries the provider's raw usage payload —
  * keyed by provider name (`{ openai: ... }`, `{ anthropic: ... }`, etc.)

package/dist/schema/events.js

@@ -39,9 +39,9 @@ import { ProviderFailureClassification } from "./errors";
  * - Anthropic and Bedrock report the input breakdown natively: Anthropic's
  *   `input_tokens` and Bedrock's `inputTokens` are non-cached only. Their
  *   mappers sum the breakdown to derive the inclusive `inputTokens`.
- *   Anthropic does *not* break extended-thinking out of `output_tokens`, so
- *   `reasoningTokens` is `undefined` and `outputTokens` carries the
- *   combined total — a documented limitation of the Anthropic API.
+ *   Anthropic's `outputTokens` includes extended thinking. Newer responses
+ *   expose that subset as `output_tokens_details.thinking_tokens`, which maps
+ *   to `reasoningTokens`; older responses leave it undefined.
  *
  * `providerMetadata` always carries the provider's raw usage payload —
  * keyed by provider name (`{ openai: ... }`, `{ anthropic: ... }`, etc.)

package/package.json

@@ -1,6 +1,6 @@
 {
   "$schema": "https://json.schemastore.org/package.json",
-  "version": "0.0.0-next-16184",
+  "version": "0.0.0-next-16187",
   "name": "@opencode-ai/ai",
   "type": "module",
   "license": "MIT",
@@ -26,7 +26,7 @@
   "devDependencies": {
     "@clack/prompts": "1.0.0-alpha.1",
     "@effect/platform-node": "4.0.0-beta.98",
-    "@opencode-ai/http-recorder": "0.0.0-next-16184",
+    "@opencode-ai/http-recorder": "0.0.0-next-16187",
     "@tsconfig/bun": "1.0.9",
     "@types/bun": "1.3.13",
     "@typescript/native-preview": "7.0.0-dev.20251207.1",
@@ -35,7 +35,7 @@
   "dependencies": {
     "@smithy/eventstream-codec": "4.2.14",
     "@smithy/util-utf8": "4.2.2",
-    "@opencode-ai/schema": "0.0.0-next-16184",
+    "@opencode-ai/schema": "0.0.0-next-16187",
     "aws4fetch": "1.0.20",
     "effect": "4.0.0-beta.98",
     "google-auth-library": "10.5.0"