chore(release): backport #33899, #33978, #34582, #34675 to rc/1.94.0 and bump litellm-proxy-extras to 0.4.79.post2

db_scripts/backfill_daily_tool_spend.sql

@@ -0,0 +1,46 @@
+-- One-shot backfill of the LiteLLM_DailyToolSpend rollup from the per-request
+-- LiteLLM_SpendLogToolIndex x LiteLLM_SpendLogs tables.
+--
+-- This is an opt-in, manual operation. New deployments do not need it: the
+-- rollup is written at request time from the moment the release is deployed.
+-- Run it only if you want the Cost Optimization "Spend by tool" card to show
+-- history from before the deploy, and only once.
+--
+-- IMPORTANT caveats before running:
+--
+-- 1. Pre-deploy index rows may include tools that were merely DECLARED in a
+--    request body but never invoked (the release this ships with stops
+--    recording those). For agentic clients that declare many tools per
+--    request, backfilled history attributes each request's full spend to
+--    every declared tool, overstating per-tool spend. Post-deploy rows do not
+--    have this problem. If your traffic is mostly such clients, consider not
+--    backfilling.
+--
+-- 2. Coverage is bounded by spend-log retention: rows older than
+--    maximum_spend_logs_retention_period are already gone.
+--
+-- 3. Replace the cutover timestamp below with the time you deployed the
+--    release, so backfilled per-request rows cannot double-count on top of
+--    rollup rows the new writer already created. ON CONFLICT DO NOTHING is a
+--    second guard for (date, tool_name) buckets the writer already touched:
+--    such buckets keep the writer's numbers and skip the backfill's.
+--
+-- Usage:
+--   psql "$DATABASE_URL" -v cutover="'2026-07-25T00:00:00Z'" -f db_scripts/backfill_daily_tool_spend.sql
+
+SET TIME ZONE 'UTC';
+
+INSERT INTO "LiteLLM_DailyToolSpend" (date, tool_name, spend, total_tokens, request_count, created_at, updated_at)
+SELECT
+    to_char(ti.start_time, 'YYYY-MM-DD') AS date,
+    ti.tool_name,
+    COALESCE(SUM(sl.spend), 0) AS spend,
+    COALESCE(SUM(sl.total_tokens), 0) AS total_tokens,
+    COUNT(*) AS request_count,
+    now() AS created_at,
+    now() AS updated_at
+FROM "LiteLLM_SpendLogToolIndex" ti
+JOIN "LiteLLM_SpendLogs" sl ON sl.request_id = ti.request_id
+WHERE ti.start_time < :cutover::timestamptz
+GROUP BY 1, 2
+ON CONFLICT (date, tool_name) DO NOTHING;

litellm-proxy-extras/litellm_proxy_extras/migrations/20260724000000_add_spend_log_tool_index_start_time_idx/migration.sql

@@ -0,0 +1,2 @@
+-- CreateIndex
+CREATE INDEX IF NOT EXISTS "LiteLLM_SpendLogToolIndex_start_time_idx" ON "LiteLLM_SpendLogToolIndex"("start_time");

litellm-proxy-extras/litellm_proxy_extras/migrations/20260725000000_add_daily_tool_spend/migration.sql

@@ -0,0 +1,12 @@
+-- CreateTable
+CREATE TABLE IF NOT EXISTS "LiteLLM_DailyToolSpend" (
+    "date" TEXT NOT NULL,
+    "tool_name" TEXT NOT NULL,
+    "spend" DOUBLE PRECISION NOT NULL DEFAULT 0.0,
+    "total_tokens" BIGINT NOT NULL DEFAULT 0,
+    "request_count" BIGINT NOT NULL DEFAULT 0,
+    "created_at" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    "updated_at" TIMESTAMP(3) NOT NULL,
+
+    CONSTRAINT "LiteLLM_DailyToolSpend_pkey" PRIMARY KEY ("date","tool_name")
+);

litellm-proxy-extras/litellm_proxy_extras/schema.prisma

@@ -1085,6 +1085,20 @@ model LiteLLM_SpendLogToolIndex {
 
   @@id([request_id, tool_name])
   @@index([tool_name, start_time])
+  @@index([start_time])
+}
+
+// Daily tool spend rollup (one row per tool per day) – the Cost Optimization card reads this, never SpendLogs
+model LiteLLM_DailyToolSpend {
+  date          String
+  tool_name     String
+  spend         Float    @default(0.0)
+  total_tokens  BigInt   @default(0)
+  request_count BigInt   @default(0)
+  created_at    DateTime @default(now())
+  updated_at    DateTime @updatedAt
+
+  @@id([date, tool_name])
 }
 
 // Prompt table for storing prompt configurations

litellm-proxy-extras/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "litellm-proxy-extras"
-version = "0.4.79.post1"
+version = "0.4.79.post2"
 description = "Additional files for the LiteLLM Proxy. Reduces the size of the main litellm package."
 readme = "README.md"
 requires-python = ">=3.9"
@@ -26,7 +26,7 @@ required-version = ">=0.10.9"
 module-root = ""
 
 [tool.commitizen]
-version = "0.4.79.post1"
+version = "0.4.79.post2"
 version_files = [
     "pyproject.toml:^version",
     "../pyproject.toml:litellm-proxy-extras==",

litellm/constants.py

@@ -1450,6 +1450,7 @@
 SPEND_LOG_CLEANUP_BATCH_FAILURE_BACKOFF_SECONDS = float(
     os.getenv("SPEND_LOG_CLEANUP_BATCH_FAILURE_BACKOFF_SECONDS", 0.5)
 )
+TOOL_SPEND_TOP_TOOLS = 100
 SPEND_LOG_PARTITION_INTERVAL = os.getenv("SPEND_LOG_PARTITION_INTERVAL", "day")
 SPEND_LOG_PARTITION_PRECREATE_AHEAD = int(os.getenv("SPEND_LOG_PARTITION_PRECREATE_AHEAD", 7))
 SPEND_LOG_QUEUE_SIZE_THRESHOLD = int(os.getenv("SPEND_LOG_QUEUE_SIZE_THRESHOLD", 100))

litellm/litellm_core_utils/prompt_templates/factory.py

@@ -6,6 +6,7 @@
 import re
 import xml.etree.ElementTree as ET
 from enum import Enum
+from collections.abc import Mapping
 from typing import Any, Dict, List, Optional, Set, Tuple, TypedDict, Union, cast, overload
 
 from jinja2.sandbox import ImmutableSandboxedEnvironment
@@ -5350,7 +5351,9 @@ def prompt_factory(
 def get_attribute_or_key(tool_or_function, attribute, default=None):
     if hasattr(tool_or_function, attribute):
         return getattr(tool_or_function, attribute)
-    return tool_or_function.get(attribute, default)
+    if isinstance(tool_or_function, Mapping):
+        return tool_or_function.get(attribute, default)
+    return default
 
 
 class NormalizedToolCall(TypedDict):
@@ -5379,14 +5382,18 @@ def _parse_tool_call_arguments(raw: Any, tool_name: Optional[str], context: str)
     return parsed if isinstance(parsed, dict) else {}
 
 
-def _tool_calls_from_chat_completion_response(response: Any) -> list[NormalizedToolCall]:
+def _tool_calls_from_chat_completion_response(
+    response: Any, include_all_choices: bool = False
+) -> list[NormalizedToolCall]:
     choices = get_attribute_or_key(response, "choices", None)
     if not (isinstance(choices, list) and choices):
         return []
-    message = get_attribute_or_key(choices[0], "message", None)
-    tool_calls = get_attribute_or_key(message, "tool_calls", None) if message else None
-    if not isinstance(tool_calls, list):
-        return []
+    tool_calls: list[Any] = []
+    for choice in choices if include_all_choices else choices[:1]:
+        message = get_attribute_or_key(choice, "message", None)
+        choice_tool_calls = get_attribute_or_key(message, "tool_calls", None) if message else None
+        if isinstance(choice_tool_calls, list):
+            tool_calls.extend(choice_tool_calls)
     result: list[NormalizedToolCall] = []
     for tc in tool_calls:
         fn = get_attribute_or_key(tc, "function", None)
@@ -5449,19 +5456,28 @@ def _tool_calls_from_anthropic_messages_response(response: Any) -> list[Normaliz
     return result
 
 
-def get_tool_calls_from_response(response: Any) -> list[NormalizedToolCall]:
+def get_tool_calls_from_response(response: Any, include_all_choices: bool = False) -> list[NormalizedToolCall]:
     """
     Extract tool/function calls from a response object into a normalized
     ``{"id", "name", "arguments"}`` shape, regardless of which API surface
     produced it: chat completions (``choices[].message.tool_calls``),
     the Responses API (``output`` items of type ``function_call``), or the
     Anthropic Messages API (``content`` blocks of type ``tool_use``).
 
+    ``include_all_choices`` decides the chat-completions scope: the default
+    reads only ``choices[0]``, which is what consumers that act on THE reply
+    (e.g. guardrails rebuilding the primary assistant message) want; usage
+    accounting passes True because every choice of an ``n>1`` request costs
+    money and its tool calls really ran. The other surfaces have a single
+    output, so the flag has no effect on them.
+
     Callers that only care about a specific tool should filter the result by
     ``name`` themselves -- this returns every tool call found.
     """
+    chat_tool_calls = _tool_calls_from_chat_completion_response(response, include_all_choices=include_all_choices)
+    if chat_tool_calls:
+        return chat_tool_calls
     for extractor in (
-        _tool_calls_from_chat_completion_response,
         _tool_calls_from_responses_api_response,
         _tool_calls_from_anthropic_messages_response,
     ):

litellm/proxy/_experimental/out/__next.!KGRhc2hib2FyZCk.__PAGE__.txt

@@ -1,9 +1,9 @@
 1:"$Sreact.fragment"
-2:I[347257,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/1uwy4yks0f1c7.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"ClientPageRoot"]
-3:I[871135,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/1uwy4yks0f1c7.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js","/litellm-asset-prefix/_next/static/chunks/1zr7rrk4wkmju.js","/litellm-asset-prefix/_next/static/chunks/4055s0-dom81s.js","/litellm-asset-prefix/_next/static/chunks/0zduf1gntl_f8.js","/litellm-asset-prefix/_next/static/chunks/20af1m4f_wo-4.js","/litellm-asset-prefix/_next/static/chunks/34z87tuwl-2tk.js","/litellm-asset-prefix/_next/static/chunks/2ovpxj_l6k1df.js","/litellm-asset-prefix/_next/static/chunks/1egl1w6n7et4v.js","/litellm-asset-prefix/_next/static/chunks/3u1stln-nfstx.js","/litellm-asset-prefix/_next/static/chunks/09il7w0a0afcs.js","/litellm-asset-prefix/_next/static/chunks/1vquuz09jxl5_.js","/litellm-asset-prefix/_next/static/chunks/04a80e3a1m-iy.js","/litellm-asset-prefix/_next/static/chunks/0dbvgsc7ha049.js","/litellm-asset-prefix/_next/static/chunks/00c27bg0wy4ve.js","/litellm-asset-prefix/_next/static/chunks/0coq3c60p1b7b.js","/litellm-asset-prefix/_next/static/chunks/3htra5tr7po1s.js","/litellm-asset-prefix/_next/static/chunks/3_6d0cxt0bjl-.js","/litellm-asset-prefix/_next/static/chunks/39dhse7b9znhd.js","/litellm-asset-prefix/_next/static/chunks/2hu1vyy-5pv13.js","/litellm-asset-prefix/_next/static/chunks/2hanmoxc51y5l.js","/litellm-asset-prefix/_next/static/chunks/0rxrzom5haz4h.js","/litellm-asset-prefix/_next/static/chunks/199uwr871eene.js","/litellm-asset-prefix/_next/static/chunks/099m1uce-fb8r.js","/litellm-asset-prefix/_next/static/chunks/1uz3jt-tj9lkf.js","/litellm-asset-prefix/_next/static/chunks/0nd2fff333ozp.js","/litellm-asset-prefix/_next/static/chunks/1agyescpxom9_.js","/litellm-asset-prefix/_next/static/chunks/296166z--6fpm.js","/litellm-asset-prefix/_next/static/chunks/12wsfsljxg4xv.js","/litellm-asset-prefix/_next/static/chunks/1y_oq-jkyyahr.js","/litellm-asset-prefix/_next/static/chunks/3-952v5v-4lvz.js","/litellm-asset-prefix/_next/static/chunks/05i7gtmth4p52.js","/litellm-asset-prefix/_next/static/chunks/23-g73xaw3kap.js","/litellm-asset-prefix/_next/static/chunks/19bxlez8idou2.js","/litellm-asset-prefix/_next/static/chunks/17arud1mz8ipv.js","/litellm-asset-prefix/_next/static/chunks/0enpuqjw6najv.js","/litellm-asset-prefix/_next/static/chunks/1h_layxqih-1d.js","/litellm-asset-prefix/_next/static/chunks/09l_m9l1emin2.js"],"default"]
-6:I[897367,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/1uwy4yks0f1c7.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"OutletBoundary"]
+2:I[347257,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/1tgw6wg8vpjb_.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"ClientPageRoot"]
+3:I[871135,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/1tgw6wg8vpjb_.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js","/litellm-asset-prefix/_next/static/chunks/1zr7rrk4wkmju.js","/litellm-asset-prefix/_next/static/chunks/4055s0-dom81s.js","/litellm-asset-prefix/_next/static/chunks/0zduf1gntl_f8.js","/litellm-asset-prefix/_next/static/chunks/04a80e3a1m-iy.js","/litellm-asset-prefix/_next/static/chunks/34z87tuwl-2tk.js","/litellm-asset-prefix/_next/static/chunks/1vquuz09jxl5_.js","/litellm-asset-prefix/_next/static/chunks/3wlz1dk9uk_5l.js","/litellm-asset-prefix/_next/static/chunks/3u1stln-nfstx.js","/litellm-asset-prefix/_next/static/chunks/09il7w0a0afcs.js","/litellm-asset-prefix/_next/static/chunks/0dbvgsc7ha049.js","/litellm-asset-prefix/_next/static/chunks/00c27bg0wy4ve.js","/litellm-asset-prefix/_next/static/chunks/0coq3c60p1b7b.js","/litellm-asset-prefix/_next/static/chunks/29zy8xbte5bx8.js","/litellm-asset-prefix/_next/static/chunks/20af1m4f_wo-4.js","/litellm-asset-prefix/_next/static/chunks/2ti12o47v07vt.js","/litellm-asset-prefix/_next/static/chunks/3_6d0cxt0bjl-.js","/litellm-asset-prefix/_next/static/chunks/2hanmoxc51y5l.js","/litellm-asset-prefix/_next/static/chunks/2hu1vyy-5pv13.js","/litellm-asset-prefix/_next/static/chunks/1uz3jt-tj9lkf.js","/litellm-asset-prefix/_next/static/chunks/0nd2fff333ozp.js","/litellm-asset-prefix/_next/static/chunks/199uwr871eene.js","/litellm-asset-prefix/_next/static/chunks/099m1uce-fb8r.js","/litellm-asset-prefix/_next/static/chunks/19bxlez8idou2.js","/litellm-asset-prefix/_next/static/chunks/0ob4xjai8e6ym.js","/litellm-asset-prefix/_next/static/chunks/09l_m9l1emin2.js","/litellm-asset-prefix/_next/static/chunks/296166z--6fpm.js","/litellm-asset-prefix/_next/static/chunks/12wsfsljxg4xv.js","/litellm-asset-prefix/_next/static/chunks/1y_oq-jkyyahr.js","/litellm-asset-prefix/_next/static/chunks/3-952v5v-4lvz.js","/litellm-asset-prefix/_next/static/chunks/05i7gtmth4p52.js","/litellm-asset-prefix/_next/static/chunks/23-g73xaw3kap.js","/litellm-asset-prefix/_next/static/chunks/2s5k4mt99snqf.js","/litellm-asset-prefix/_next/static/chunks/17arud1mz8ipv.js","/litellm-asset-prefix/_next/static/chunks/10hqekmjsynkf.js","/litellm-asset-prefix/_next/static/chunks/1h_layxqih-1d.js"],"default"]
+6:I[897367,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/1tgw6wg8vpjb_.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"OutletBoundary"]
 7:"$Sreact.suspense"
-0:{"rsc":["$","$1","c",{"children":[["$","$L2",null,{"Component":"$3","serverProvidedParams":{"searchParams":{},"params":{},"promises":["$@4","$@5"]}}],[["$","script","script-0",{"src":"/litellm-asset-prefix/_next/static/chunks/3_6d0cxt0bjl-.js","async":true}],["$","script","script-1",{"src":"/litellm-asset-prefix/_next/static/chunks/39dhse7b9znhd.js","async":true}],["$","script","script-2",{"src":"/litellm-asset-prefix/_next/static/chunks/2hu1vyy-5pv13.js","async":true}],["$","script","script-3",{"src":"/litellm-asset-prefix/_next/static/chunks/2hanmoxc51y5l.js","async":true}],["$","script","script-4",{"src":"/litellm-asset-prefix/_next/static/chunks/0rxrzom5haz4h.js","async":true}],["$","script","script-5",{"src":"/litellm-asset-prefix/_next/static/chunks/199uwr871eene.js","async":true}],["$","script","script-6",{"src":"/litellm-asset-prefix/_next/static/chunks/099m1uce-fb8r.js","async":true}],["$","script","script-7",{"src":"/litellm-asset-prefix/_next/static/chunks/1uz3jt-tj9lkf.js","async":true}],["$","script","script-8",{"src":"/litellm-asset-prefix/_next/static/chunks/0nd2fff333ozp.js","async":true}],["$","script","script-9",{"src":"/litellm-asset-prefix/_next/static/chunks/1agyescpxom9_.js","async":true}],["$","script","script-10",{"src":"/litellm-asset-prefix/_next/static/chunks/296166z--6fpm.js","async":true}],["$","script","script-11",{"src":"/litellm-asset-prefix/_next/static/chunks/12wsfsljxg4xv.js","async":true}],["$","script","script-12",{"src":"/litellm-asset-prefix/_next/static/chunks/1y_oq-jkyyahr.js","async":true}],["$","script","script-13",{"src":"/litellm-asset-prefix/_next/static/chunks/3-952v5v-4lvz.js","async":true}],["$","script","script-14",{"src":"/litellm-asset-prefix/_next/static/chunks/05i7gtmth4p52.js","async":true}],["$","script","script-15",{"src":"/litellm-asset-prefix/_next/static/chunks/23-g73xaw3kap.js","async":true}],["$","script","script-16",{"src":"/litellm-asset-prefix/_next/static/chunks/19bxlez8idou2.js","async":true}],["$","script","script-17",{"src":"/litellm-asset-prefix/_next/static/chunks/17arud1mz8ipv.js","async":true}],["$","script","script-18",{"src":"/litellm-asset-prefix/_next/static/chunks/0enpuqjw6najv.js","async":true}],["$","script","script-19",{"src":"/litellm-asset-prefix/_next/static/chunks/1h_layxqih-1d.js","async":true}],["$","script","script-20",{"src":"/litellm-asset-prefix/_next/static/chunks/09l_m9l1emin2.js","async":true}]],["$","$L6",null,{"children":["$","$7",null,{"name":"Next.MetadataOutlet","children":"$@8"}]}]]}],"isPartial":false,"staleTime":300,"varyParams":null,"buildId":"tjyp94vnPkfmefK0pvYUs"}
+0:{"rsc":["$","$1","c",{"children":[["$","$L2",null,{"Component":"$3","serverProvidedParams":{"searchParams":{},"params":{},"promises":["$@4","$@5"]}}],[["$","script","script-0",{"src":"/litellm-asset-prefix/_next/static/chunks/3_6d0cxt0bjl-.js","async":true}],["$","script","script-1",{"src":"/litellm-asset-prefix/_next/static/chunks/2hanmoxc51y5l.js","async":true}],["$","script","script-2",{"src":"/litellm-asset-prefix/_next/static/chunks/2hu1vyy-5pv13.js","async":true}],["$","script","script-3",{"src":"/litellm-asset-prefix/_next/static/chunks/1uz3jt-tj9lkf.js","async":true}],["$","script","script-4",{"src":"/litellm-asset-prefix/_next/static/chunks/0nd2fff333ozp.js","async":true}],["$","script","script-5",{"src":"/litellm-asset-prefix/_next/static/chunks/199uwr871eene.js","async":true}],["$","script","script-6",{"src":"/litellm-asset-prefix/_next/static/chunks/099m1uce-fb8r.js","async":true}],["$","script","script-7",{"src":"/litellm-asset-prefix/_next/static/chunks/19bxlez8idou2.js","async":true}],["$","script","script-8",{"src":"/litellm-asset-prefix/_next/static/chunks/0ob4xjai8e6ym.js","async":true}],["$","script","script-9",{"src":"/litellm-asset-prefix/_next/static/chunks/09l_m9l1emin2.js","async":true}],["$","script","script-10",{"src":"/litellm-asset-prefix/_next/static/chunks/296166z--6fpm.js","async":true}],["$","script","script-11",{"src":"/litellm-asset-prefix/_next/static/chunks/12wsfsljxg4xv.js","async":true}],["$","script","script-12",{"src":"/litellm-asset-prefix/_next/static/chunks/1y_oq-jkyyahr.js","async":true}],["$","script","script-13",{"src":"/litellm-asset-prefix/_next/static/chunks/3-952v5v-4lvz.js","async":true}],["$","script","script-14",{"src":"/litellm-asset-prefix/_next/static/chunks/05i7gtmth4p52.js","async":true}],["$","script","script-15",{"src":"/litellm-asset-prefix/_next/static/chunks/23-g73xaw3kap.js","async":true}],["$","script","script-16",{"src":"/litellm-asset-prefix/_next/static/chunks/2s5k4mt99snqf.js","async":true}],["$","script","script-17",{"src":"/litellm-asset-prefix/_next/static/chunks/17arud1mz8ipv.js","async":true}],["$","script","script-18",{"src":"/litellm-asset-prefix/_next/static/chunks/10hqekmjsynkf.js","async":true}],["$","script","script-19",{"src":"/litellm-asset-prefix/_next/static/chunks/1h_layxqih-1d.js","async":true}]],["$","$L6",null,{"children":["$","$7",null,{"name":"Next.MetadataOutlet","children":"$@8"}]}]]}],"isPartial":false,"staleTime":300,"varyParams":null,"buildId":"42XogNvS9fAzsVum7mqhy"}
 4:{}
 5:"$0:rsc:props:children:0:props:serverProvidedParams:params"
 8:null