Benchmark Almanac

PMBench

KARL 49.0 tops its leaderboard Claude Opus 4.6 46.1 leads the models we track
Not in model cards

Agents search product managers’ notes to identify evidence across customer conversations and answer questions about product requirements.

The report compares seven GPT and Claude models on 57 questions over roughly 3,000 internal documents.

Released
March 2026
Kind
Questions that need search
Level
University
Scored
Graded by a model
Answers
Short answer
Tasks
Kept private
Measures
Capability
Language
English
In and out
Documents, Text → Text
Interaction
Single turn
Items from
Real work and use
Refresh
One fixed set
Occupations
Management, Business and financial operations
Strongest model tested
Claude Opus 4.6 #39 on the capability index
Last model added
February 2026
Scores refreshed
November 21st, 2026
13
Entries in its paper, 10 models we track

KARL tops the leaderboard with 49.0 Claude Opus 4.6 leads with 46.1 among models released by February 2026

Best score per entry from the benchmark's paper, read on November 21st, 2026. A paper’s table doesn’t add newer models. 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. The best model we track is Claude Opus 4.6, at 46.1. This view shows only models we track, each at its best.

The newest model with a score here came out in February 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 KARL parallel N = 20 49.0
  2. 2 Claude Opus 4.6 46.1
  3. 3 Claude Sonnet 4.6 43.8
  4. 4 Claude Opus 4.5 39.1
  5. 5 GPT-5.2 37.9
  6. 6 KARL-TREC 37.5
  7. 7 GPT-5 37.5
  8. 8 Claude Haiku 4.5 35.0
  9. 9 Qwen3.5 397B-A17B 34.7
  10. 10 MiniMax-M2.5 34.5
  11. 11 GLM-4.5-Air 33.4
  12. 12 Claude Sonnet 4.5 32.6
  13. 13 KARL-BCP 32.4

Source: The PMBench paper

#EntryScoreSettingDateSource
1 KARL 49.0 parallel N = 20 Official leaderboard
2 Claude Opus 4.6 46.1 Feb 2026 Official leaderboard
3 Claude Sonnet 4.6 43.8 Feb 2026 Official leaderboard
4 Claude Opus 4.5 39.1 Nov 2025 Official leaderboard
5 GPT-5.2 37.9 Dec 2025 Official leaderboard
6 KARL-TREC 37.5 Official leaderboard
7 GPT-5 37.5 Aug 2025 Official leaderboard
8 Claude Haiku 4.5 35.0 Oct 2025 Official leaderboard
9 Qwen3.5 397B-A17B 34.7 Feb 2026 Official leaderboard
10 MiniMax-M2.5 34.5 Feb 2026 Official leaderboard
11 GLM-4.5-Air 33.4 Jul 2025 Official leaderboard
12 Claude Sonnet 4.5 32.6 Sep 2025 Official leaderboard
13 KARL-BCP 32.4 Official leaderboard
#ModelScoreSettingDateSource
1 Claude Opus 4.6 46.1 Feb 2026 Official leaderboard
2 Claude Sonnet 4.6 43.8 Feb 2026 Official leaderboard
3 Claude Opus 4.5 39.1 Nov 2025 Official leaderboard
4 GPT-5.2 37.9 Dec 2025 Official leaderboard
5 GPT-5 37.5 Aug 2025 Official leaderboard
6 Claude Haiku 4.5 35.0 Oct 2025 Official leaderboard
7 Qwen3.5 397B-A17B 34.7 Feb 2026 Official leaderboard
8 MiniMax-M2.5 34.5 Feb 2026 Official leaderboard
9 GLM-4.5-Air 33.4 Jul 2025 Official leaderboard
10 Claude Sonnet 4.5 32.6 Sep 2025 Official leaderboard

Number one over time

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

Anthropic
Benchmark paper Not tested since Mar 2026

Source: The PMBench paper

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
In the benchmark’s paper
30.0 35.0 40.0 45.0 50.0 2026 30.0 35.0 40.0 45.0 50.0 2026

Source: Official PMBench 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 Knowledge

All 332 in Knowledge
  • MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects. Intern-S2-397B 89.8%
  • MMLU In use 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy. Seed1.8 92.3%
  • SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy. Qwen 3.6 Max (Preview) 73.9%
  • WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1. Tencent Hy4 preview 83.9%
  • BBEH micro average In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is the share of all examples answered correctly. Gemma 4 31B IT 74.4%
  • HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. Claude Fable 5.2 77.9%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 9 months ago, in February 2026, for Claude Sonnet 4.6. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the benchmark’s paper (13 entries).
  • Pass Models scored 10 models we track, plus 3 other entries on its leaderboard.
  • Warning Newest score The newest came 9 months ago, in February 2026, for Claude Sonnet 4.6.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude Opus 4.6 (#39).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on databricks.com: 10 rows for models we track and 3 other entries at the last read.
  • Pass Last read November 21st, 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