Shopping MMLU
Models complete 57 shopping tasks involving product concepts, reasoning, user preferences, and several languages. Task-specific metrics check classification, ranking, extraction, translation, and generated text. Skill scores average the corresponding task metrics.
The suite contains 20,799 questions in six languages and draws on public, internal, and synthetic Amazon data. Its aggregate combines different metrics. It evaluates text-based shopping assistance rather than browser navigation.
- Made by
- Amazon and academic collaborators
- Released
- October 2024
- Kind
- Running a business
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Language
- Many languages
- Interaction
- Single turn
- Items from
- Real work and use, Generated
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Many languages
- Strongest model tested
-
Llama 3-70B #182 on the capability index
- Last model added
- April 2024
- Scores refreshed
- November 18th, 2026
In the authors’ words
Online shopping is a complex multi-task, few-shot learning problem with a wide and evolving range of entities, relations, and tasks. However, existing models and benchmarks are commonly tailored to specific tasks, falling short of capturing the full complexity of online shopping. Large Language Models (LLMs), with their multi-task and few-shot learning abilities, have the potential to profoundly transform online shopping by alleviating task-specific engineering efforts and by providing users with interactive conversations. Despite the potential, LLMs face unique challenges in online shopping, such as domain-specific concepts, implicit knowledge, and heterogeneous user behaviors. Motivated by the potential and challenges, we propose Shopping MMLU, a diverse multi-task online shopping benchmark derived from real-world Amazon data. Shopping MMLU consists of 57 tasks covering 4 major shopping skills: concept understanding, knowledge reasoning, user behavior alignment, and multi-linguality, and can thus comprehensively evaluate the abilities of LLMs as general shop assistants. With Shopping MMLU, we benchmark over 20 existing LLMs and uncover valuable insights about practices and prospects of building versatile LLM-based shop assistants. Shopping MMLU can be publicly accessed at https://github.com/KL4805/ShoppingMMLU. In addition, with Shopping MMLU, we host a competition in KDD Cup 2024 with over 500 participating teams. The winning solutions and the associated workshop can be accessed at our website https://amazon-kddcup24.github.io/.
Claude 3 Sonnet leads with 72.6 among models released by April 2024 Claude 3 Sonnet leads with 72.6 among models released by April 2024
Best score per entry from the benchmark's own leaderboard, read on November 18th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in April 2024, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 72.6 Claude 3 Sonnet
- 68.6 LLaMA3-70B-Instruct
- 67.4 Claude 2
- 67.4
Qwen1.5-72B
- 65.3 GPT-3.5-Turbo (snapshot unspecified)
- 62.0
Llama 3-70B
- 59.6
Qwen1.5-14B
- 58.1 LLaMA3-8B-Instruct
- 57.6 eCeLLM-L
- 55.5 eCeLLM-M
- 53.4
Mixtral 8x7B
- 52.9 Vicuna-13B-v1.5
- 52.8
Qwen1.5-7B
- 52.0
Llama 2-70B
- 51.1 Zephyr-Beta
- 50.7
Llama 3-8B
- 50.6 QWen1.5-4B
- 48.5 LLaMA2-70B-Chat
- 48.5
Mistral-7B-Instruct-v0.1
- 47.6
Mistral 7B v0.1
- 45.9 Vicuna-7B-v1.5
- 44.9 LLaMA2-13B-Chat
- 44.3 LLaMA2-7B-Chat
- 42.2
Llama 2-13B
- 40.4
Phi-2
- 39.4 eCeLLM-S
- 32.8
Llama 2-7B
- 1 Claude 3 Sonnet 72.6
- 2 LLaMA3-70B-Instruct 68.6
- 3 Claude 2 67.4
- 4
Qwen1.5-72B 67.4
- 5 GPT-3.5-Turbo (snapshot unspecified) 65.3
- 6
Llama 3-70B 62.0
- 7
Qwen1.5-14B 59.6
- 8 LLaMA3-8B-Instruct 58.1
- 9 eCeLLM-L 57.6
- 10 eCeLLM-M 55.5
- 11
Mixtral 8x7B 53.4
- 12 Vicuna-13B-v1.5 52.9
- 13
Qwen1.5-7B 52.8
- 14
Llama 2-70B 52.0
- 15 Zephyr-Beta 51.1
- 16
Llama 3-8B 50.7
- 17 QWen1.5-4B 50.6
- 18 LLaMA2-70B-Chat 48.5
- 19
Mistral-7B-Instruct-v0.1 48.5
- 20
Mistral 7B v0.1 47.6
- 21 Vicuna-7B-v1.5 45.9
- 22 LLaMA2-13B-Chat 44.9
- 23 LLaMA2-7B-Chat 44.3
- 24
Llama 2-13B 42.2
- 25
Phi-2 40.4
- 26 eCeLLM-S 39.4
- 27
Llama 2-7B 32.8
Source: Official Shopping MMLU leaderboard
- 1 Claude 3 Sonnet 72.6
- 2 Claude 2 67.4
- 3
Qwen1.5-72B 67.4
- 4
Llama 3-70B 62.0
- 5
Qwen1.5-14B 59.6
- 6
Mixtral 8x7B 53.4
- 7
Qwen1.5-7B 52.8
- 8
Llama 2-70B 52.0
- 9
Llama 3-8B 50.7
- 10
Mistral-7B-Instruct-v0.1 48.5
- 11
Mistral 7B v0.1 47.6
- 12
Llama 2-13B 42.2
- 13
Phi-2 40.4
- 14
Llama 2-7B 32.8
Source: Official Shopping MMLU leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude 3 Sonnet | 72.6 | Feb 2024 | Official leaderboard | |
| 2 | LLaMA3-70B-Instruct | 68.6 | Official leaderboard | ||
| 3 | Claude 2 | 67.4 | Jul 2023 | Official leaderboard | |
| 4 | | 67.4 | Feb 2024 | Official leaderboard | |
| 5 | GPT-3.5-Turbo (snapshot unspecified) | 65.3 | Official leaderboard | ||
| 6 | | 62.0 | Apr 2024 | Official leaderboard | |
| 7 | | 59.6 | Feb 2024 | Official leaderboard | |
| 8 | LLaMA3-8B-Instruct | 58.1 | Official leaderboard | ||
| 9 | eCeLLM-L | 57.6 | Official leaderboard | ||
| 10 | eCeLLM-M | 55.5 | Official leaderboard | ||
| 11 | | 53.4 | Dec 2023 | Official leaderboard | |
| 12 | Vicuna-13B-v1.5 | 52.9 | Official leaderboard | ||
| 13 | | 52.8 | Feb 2024 | Official leaderboard | |
| 14 | | 52.0 | Jul 2023 | Official leaderboard | |
| 15 | Zephyr-Beta | 51.1 | Official leaderboard | ||
| 16 | | 50.7 | Apr 2024 | Official leaderboard | |
| 17 | QWen1.5-4B | 50.6 | Official leaderboard | ||
| 18 | LLaMA2-70B-Chat | 48.5 | Official leaderboard | ||
| 19 | | 48.5 | Sep 2023 | Official leaderboard | |
| 20 | | 47.6 | Sep 2023 | Official leaderboard | |
| 21 | Vicuna-7B-v1.5 | 45.9 | Official leaderboard | ||
| 22 | LLaMA2-13B-Chat | 44.9 | Official leaderboard | ||
| 23 | LLaMA2-7B-Chat | 44.3 | Official leaderboard | ||
| 24 | | 42.2 | Jul 2023 | Official leaderboard | |
| 25 | | 40.4 | Dec 2023 | Official leaderboard | |
| 26 | eCeLLM-S | 39.4 | Official leaderboard | ||
| 27 | | 32.8 | Jul 2023 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude 3 Sonnet | 72.6 | Feb 2024 | Official leaderboard | |
| 2 | Claude 2 | 67.4 | Jul 2023 | Official leaderboard | |
| 3 | | 67.4 | Feb 2024 | Official leaderboard | |
| 4 | | 62.0 | Apr 2024 | Official leaderboard | |
| 5 | | 59.6 | Feb 2024 | Official leaderboard | |
| 6 | | 53.4 | Dec 2023 | Official leaderboard | |
| 7 | | 52.8 | Feb 2024 | Official leaderboard | |
| 8 | | 52.0 | Jul 2023 | Official leaderboard | |
| 9 | | 50.7 | Apr 2024 | Official leaderboard | |
| 10 | | 48.5 | Sep 2023 | Official leaderboard | |
| 11 | | 47.6 | Sep 2023 | Official leaderboard | |
| 12 | | 42.2 | Jul 2023 | Official leaderboard | |
| 13 | | 40.4 | Dec 2023 | Official leaderboard | |
| 14 | | 32.8 | Jul 2023 | Official leaderboard |
Rankings by part
We read 4 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Shopping concept understanding, Shopping knowledge reasoning and User behavior alignment. Claude 3 Sonnet leads every one of them. Arrows show how far a model moved from its overall rank among the models we track.
Shopping concept understanding
14 models we track have a score here.
- 1 Claude 3 Sonnet 80.8
- 2 Claude 2 75.5
- 3
Qwen1.5-72B 71.7
- 4
Llama 3-70B 69.6
- 5
Qwen1.5-14B 67.2
- 6
Mistral-7B-Instruct-v0.1 ↑4 62.0
- 7
Llama 2-70B ↑1 61.0
- 8
Mixtral 8x7B ↓2 59.4
- 9
Qwen1.5-7B ↓2 58.9
- 10
Llama 3-8B ↓1 58.0
Shopping knowledge reasoning
14 models we track have a score here.
- 1 Claude 3 Sonnet 71.6
- 2
Qwen1.5-72B ↑1 68.9
- 3 Claude 2 ↓1 65.5
- 4
Llama 3-70B 63.6
- 5
Qwen1.5-14B 60.9
- 6
Llama 2-70B ↑2 55.9
- 7
Mixtral 8x7B ↓1 54.3
- 8
Qwen1.5-7B ↓1 52.3
- 9
Llama 3-8B 49.7
- 10
Mistral 7B v0.1 ↑1 46.7
User behavior alignment
14 models we track have a score here.
- 1 Claude 3 Sonnet 70.2
- 2
Qwen1.5-72B ↑1 64.1
- 3 Claude 2 ↓1 63.5
- 4
Llama 3-70B 55.8
- 5
Mixtral 8x7B ↑1 55.3
- 6
Qwen1.5-14B ↓1 54.9
- 7
Qwen1.5-7B 49.8
- 8
Mistral 7B v0.1 ↑3 46.3
- 9
Llama 3-8B 44.2
- 10
Llama 2-70B ↓2 43.2
Multilingual abilities
14 models we track have a score here.
- 1 Claude 3 Sonnet 67.8
- 2 Claude 2 65.2
- 3
Qwen1.5-72B 64.8
- 4
Llama 3-70B 59.0
- 5
Qwen1.5-14B 55.2
- 6
Llama 3-8B ↑3 51.0
- 7
Qwen1.5-7B 50.1
- 8
Llama 2-70B 47.9
- 9
Mixtral 8x7B ↓3 44.7
- 10
Llama 2-13B ↑2 44.2
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official Shopping MMLU leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official Shopping MMLU leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Economic work
- JobBench In use 130 tasks from 35 occupations workers most want delegated. Agent works through messy reference files, graded by fact-anchored binary rubrics. GLM-5.3 61.4%
- GDPval In use Real work tasks from 44 occupations with reference files. Experts blindly compare model deliverables with expert work. Metric is win-or-tie rate.
Seed2.1 Pro 87.9%
- GDPval-AA v2.1 In use Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Blind pairwise comparisons give Elo ratings. Claude Opus 5.5 1,846
- AA-Briefcase v1.1 In use Version 1.1 of Artificial Analysis's multi-hour office work evaluation, with Elo ratings fitted by Crowd-BT. Claude Opus 5.5 1,822
- Agents' Last Exam v1.1 In use Agents complete 122 public long-horizon professional workflow tasks across 55 sub-industries. Deterministic graders check outcomes against hidden references. GPT-6.2 Astra 47.5%
- CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished). Claude Fable 5 75.9%
Data audit
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 2 years ago, in April 2024, for Llama 3-70B. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (27 entries).
- Pass Models scored 14 models we track, plus 13 other entries on its leaderboard.
- Warning Newest score The newest came 2 years ago, in April 2024, for Llama 3-70B.
- Note Frontier models None of the top 10 has a score. The strongest tested is Llama 3-70B (#182).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on huggingface.co: 14 rows for models we track and 13 other entries at the last read.
- Pass Last read November 18th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.