Benchmark Almanac

Shopping MMLU

Claude 3 Sonnet 72.6 leads since October 2024
Not in model cards

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.

Released
October 2024
Kind
Running a business
Level
Everyday
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
Many languages
In and out
Text → Text
Interaction
Single turn
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/.

Abstract of Shopping MMLU: A Massive Multi-Task Online Shopping Benchmark for Large Language Models (arXiv:2410.20745, CC0)
27
Entries on its own leaderboard, 14 models we track

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:

  1. 1 Claude 3 Sonnet 72.6
  2. 2 LLaMA3-70B-Instruct 68.6
  3. 3 Claude 2 67.4
  4. 4 Qwen1.5-72B 67.4
  5. 5 GPT-3.5-Turbo (snapshot unspecified) 65.3
  6. 6 Llama 3-70B 62.0
  7. 7 Qwen1.5-14B 59.6
  8. 8 LLaMA3-8B-Instruct 58.1
  9. 9 eCeLLM-L 57.6
  10. 10 eCeLLM-M 55.5
  11. 11 Mixtral 8x7B 53.4
  12. 12 Vicuna-13B-v1.5 52.9
  13. 13 Qwen1.5-7B 52.8
  14. 14 Llama 2-70B 52.0
  15. 15 Zephyr-Beta 51.1
  16. 16 Llama 3-8B 50.7
  17. 17 QWen1.5-4B 50.6
  18. 18 LLaMA2-70B-Chat 48.5
  19. 19 Mistral-7B-Instruct-v0.1 48.5
  20. 20 Mistral 7B v0.1 47.6
  21. 21 Vicuna-7B-v1.5 45.9
  22. 22 LLaMA2-13B-Chat 44.9
  23. 23 LLaMA2-7B-Chat 44.3
  24. 24 Llama 2-13B 42.2
  25. 25 Phi-2 40.4
  26. 26 eCeLLM-S 39.4
  27. 27 Llama 2-7B 32.8

Source: Official Shopping MMLU leaderboard

#EntryScoreSettingDateSource
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 Qwen1.5-72B 67.4 Feb 2024 Official leaderboard
5 GPT-3.5-Turbo (snapshot unspecified) 65.3 Official leaderboard
6 Llama 3-70B 62.0 Apr 2024 Official leaderboard
7 Qwen1.5-14B 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 Mixtral 8x7B 53.4 Dec 2023 Official leaderboard
12 Vicuna-13B-v1.5 52.9 Official leaderboard
13 Qwen1.5-7B 52.8 Feb 2024 Official leaderboard
14 Llama 2-70B 52.0 Jul 2023 Official leaderboard
15 Zephyr-Beta 51.1 Official leaderboard
16 Llama 3-8B 50.7 Apr 2024 Official leaderboard
17 QWen1.5-4B 50.6 Official leaderboard
18 LLaMA2-70B-Chat 48.5 Official leaderboard
19 Mistral-7B-Instruct-v0.1 48.5 Sep 2023 Official leaderboard
20 Mistral 7B v0.1 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 Llama 2-13B 42.2 Jul 2023 Official leaderboard
25 Phi-2 40.4 Dec 2023 Official leaderboard
26 eCeLLM-S 39.4 Official leaderboard
27 Llama 2-7B 32.8 Jul 2023 Official leaderboard
#ModelScoreSettingDateSource
1 Claude 3 Sonnet 72.6 Feb 2024 Official leaderboard
2 Claude 2 67.4 Jul 2023 Official leaderboard
3 Qwen1.5-72B 67.4 Feb 2024 Official leaderboard
4 Llama 3-70B 62.0 Apr 2024 Official leaderboard
5 Qwen1.5-14B 59.6 Feb 2024 Official leaderboard
6 Mixtral 8x7B 53.4 Dec 2023 Official leaderboard
7 Qwen1.5-7B 52.8 Feb 2024 Official leaderboard
8 Llama 2-70B 52.0 Jul 2023 Official leaderboard
9 Llama 3-8B 50.7 Apr 2024 Official leaderboard
10 Mistral-7B-Instruct-v0.1 48.5 Sep 2023 Official leaderboard
11 Mistral 7B v0.1 47.6 Sep 2023 Official leaderboard
12 Llama 2-13B 42.2 Jul 2023 Official leaderboard
13 Phi-2 40.4 Dec 2023 Official leaderboard
14 Llama 2-7B 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. 1 Claude 3 Sonnet 80.8
  2. 2 Claude 2 75.5
  3. 3 Qwen1.5-72B 71.7
  4. 4 Llama 3-70B 69.6
  5. 5 Qwen1.5-14B 67.2
  6. 6 Mistral-7B-Instruct-v0.1 ↑4 62.0
  7. 7 Llama 2-70B ↑1 61.0
  8. 8 Mixtral 8x7B ↓2 59.4
  9. 9 Qwen1.5-7B ↓2 58.9
  10. 10 Llama 3-8B ↓1 58.0

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic
Own leaderboard Not tested since Oct 2024

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.

Plot with other benchmarks
On the benchmark’s own leaderboard
30.0 40.0 50.0 60.0 70.0 80.0 2024 30.0 40.0 50.0 60.0 70.0 80.0 2024

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

All 210 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.

How we check