SAGE (Service Agent) - Logistics Delivery: leaderboard

SAGE overall assessment score in the logistics delivery customer-service scenario: logistics exceptions such as lost or delayed parcels and insurance claims, combining procedure compliance and conversational quality on a 0-100 scale, higher is better. It is unrelated to the SAGE robustness benchmark.

Metric: Overall Assessment Score (0-100), 0.8 x logical compliance + 0.2 conversational quality, in the Logistics Delivery scenario (logistics exceptions such as lost or delayed parcels and insurance claims), over SAGE customer-service dialogues: standard operating procedures formalized as dialogue graphs, an adversarial simulated user with varied intents and personas, turns 1, 5, 10, 15 and the final turn scored; three unnamed judge LLMs (majority vote and mean) plus a deterministic rule engine produce the reference; closed models through their APIs, open models served with vLLM; 0-100 scale; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 27 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Claude Opus 4.571.25#160
2DeepSeek V3.270.87#419
3MiniMax-M2.168.54#575
4Gemini 3 Pro (Preview)68.05#119
5Claude Sonnet 4.567.3#261
6Gemini 3 Flash (Preview)67.1#152
7GLM-4.766.5#380
8DeepSeek R165.75#482
9Kimi K2.565.74#266
10Gemini 2.5 Pro64.92#276
11Qwen 3 235B A22B64.77#628
12Qwen Max (SAGE checkpoint unspecified)64.03
13Seed 1.864.01#252
14Qwen 2.5 72B Instruct63.16#1117
15Qwen 2.5 32B Instruct62.99#1247

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=sage-service-agent-logistics-delivery · How It Works · Data refreshed daily, snapshot 2026-11-21.