Benchmark Almanac

JFBench

PLaMo 3.0 Prime 60.6% tops its leaderboard Muse Glimmer 48.9% leads the models we track
Not in model cards

Models answer Japanese prompts with multiple verifiable constraints. A response passes only if it satisfies every constraint.

The verified evaluation reports pass rates by the number of constraints per prompt.

Released
March 2026
Kind
Chat and instructions in many languages
Level
Everyday
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
Japanese
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Strongest model tested
Qwen 3.8 27B #59 on the capability index
Last model added
August 2026
Scores refreshed
November 21st, 2026
10
Entries on its own leaderboard, 6 models we track

PLaMo 3.0 Prime tops the leaderboard with 60.6% Muse Glimmer leads with 48.9% among models released by August 2026

Best score per entry from the benchmark's own leaderboard, read on November 21st, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. 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 Muse Glimmer, at 48.9%. This view shows only models we track, each at its best.

The newest model with a score here came out in August 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 PLaMo 3.0 Prime reasoning 60.6%
  2. 2 PLaMo 3.1 Prime reasoning 55.1%
  3. 3 ELYZA-Thinking 33B reasoning 51.8%
  4. 4 Muse Glimmer reasoning 48.9%
  5. 5 Gemma 4 31B IT reasoning 48.5%
  6. 6 Qwen3.6 27B reasoning 48.1%
  7. 7 Qwen 3.8 27B reasoning 43.5%
  8. 8 llm-jp-4.1 33B Thinking reasoning 39.8%
  9. 9 gpt-oss-120b reasoning 39.6%
  10. 10 Gemma 3 27B non-reasoning 25.8%

Source: Official JFBench leaderboard

#EntryScoreSettingDateSource
1 PLaMo 3.0 Prime 60.6% reasoning Official leaderboard
2 PLaMo 3.1 Prime 55.1% reasoning Official leaderboard
3 ELYZA-Thinking 33B 51.8% reasoning Official leaderboard
4 Muse Glimmer 48.9% reasoning Aug 2026 Official leaderboard
5 Gemma 4 31B IT 48.5% reasoning Apr 2026 Official leaderboard
6 Qwen3.6 27B 48.1% reasoning Apr 2026 Official leaderboard
7 Qwen 3.8 27B 43.5% reasoning Aug 2026 Official leaderboard
8 llm-jp-4.1 33B Thinking 39.8% reasoning Official leaderboard
9 gpt-oss-120b 39.6% reasoning Aug 2025 Official leaderboard
10 Gemma 3 27B 25.8% non-reasoning Mar 2025 Official leaderboard
#ModelScoreSettingDateSource
1 Muse Glimmer 48.9% reasoning Aug 2026 Official leaderboard
2 Gemma 4 31B IT 48.5% reasoning Apr 2026 Official leaderboard
3 Qwen3.6 27B 48.1% reasoning Apr 2026 Official leaderboard
4 Qwen 3.8 27B 43.5% reasoning Aug 2026 Official leaderboard
5 gpt-oss-120b 39.6% reasoning Aug 2025 Official leaderboard
6 Gemma 3 27B 25.8% non-reasoning Mar 2025 Official leaderboard

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Gemma 4 31B IT
non-reasoni… reasoning
48.5%
Qwen3.6 27B
non-reasoni… reasoning
48.1%
Qwen 3.8 27B
non-reasoni… reasoning
43.5%

Source: Official JFBench leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Muse Glimmer reasoning 48.9%
Gemma 4 31B IT reasoning 48.5%
Qwen3.6 27B reasoning 48.1%
Qwen 3.8 27B reasoning 43.5%
gpt-oss-120b reasoning 39.6%
Gemma 4 31B IT non-reasoning 38.6%
Qwen 3.8 27B non-reasoning 33.8%
Qwen3.6 27B non-reasoning 30.9%
Gemma 3 27B non-reasoning 25.8%

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Gemma 4 31B IT
non-reasoni… reasoning
48.5%
Qwen3.6 27B
non-reasoni… reasoning
48.1%
Qwen 3.8 27B
non-reasoni… reasoning
43.5%

Source: Official JFBench leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Muse Glimmer reasoning 48.9%
Gemma 4 31B IT reasoning 48.5%
Qwen3.6 27B reasoning 48.1%
Qwen 3.8 27B reasoning 43.5%
gpt-oss-120b reasoning 39.6%
Gemma 4 31B IT non-reasoning 38.6%
Qwen 3.8 27B non-reasoning 33.8%
Qwen3.6 27B non-reasoning 30.9%
Gemma 3 27B non-reasoning 25.8%

3 models have held number one on the official leaderboard since its release in March 2026

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

OpenAI Google DeepMind Meta
Own leaderboard Not tested since Aug 2026
Muse Glimmer 48.9%

Source: Official JFBench 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 Best so far
0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Muse Glimmer Gemma 4 31B IT gpt-oss-120b Gemma 3 27B 0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Muse Glimmer gpt-oss-120b

Source: Official JFBench 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 Multilingual

All 545 in Multilingual
  • MMMLU In use MMLU professionally translated into 14 languages. Multiple-choice accuracy. Claude Mythos Preview 92.7%
  • WMT24++ In use English-to-55-language translation of WMT24 test data. Scored with automatic metrics such as MetricX or xCOMET. K-EXAONE 90.5%
  • MMLU-ProX In use MMLU-Pro translated into up to 29 languages with expert review. Multiple-choice accuracy per language. Qwen3.7-Max 87%
  • CLIcK In use Models answer 1,995 multiple-choice questions on Korean language and culture. Questions come from official Korean exams and textbooks in eleven categories. Solar Open 2 90.7%
  • Global MMLU Lite 3.0 In use Models answer 400 human-translated MMLU questions per language in 23 languages. Version 3.0 adds Czech, Hungarian, Oriya, Slovak and Tajik and updates Italian. Accuracy. Claude Opus 5 95.7%
  • MMMB In use Models answer multiple-choice questions about images in six languages across 15 categories. The set has about 12,000 questions. LFM2.5-VL-3B 83%

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (10 entries).
  • Pass Models scored 6 models we track, plus 4 other entries on its leaderboard.
  • Pass Newest score The newest is from August 2026, for Qwen 3.8 27B.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Qwen 3.8 27B (#59).
  • Pass Headroom The best is 48.9%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 9 rows for models we track and 4 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