LiT (Lost in Translation) - Humanities Abstracts: leaderboard

LiT, Lost in Translation (2026), measures multilingual proficiency by round-trip translation: English passages pass through a sequence of four languages and back, and an LLM judge scores the result against the source with MQM error weights. Share of humanities paper abstracts whose round trip scores at least 80, in percent; higher is better.

Metric: MQM>=80 rate (%): share of round-trip translations whose back-translation to English scores at least 80 on an MQM error scale (minor -1, major -5, critical -25) from a Grok 4.1 Fast judge comparing it with the English source; each passage is translated serially through a four-language sequence and back, averaged over eight sequences (high, medium and low-resource languages); humanities paper abstracts; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 17 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3 FlashNon-reasoning90#175
2Gemma 4 31B (IT)71.2#350
3Qwen 3.5 397B A17BNon-reasoning71.2#264
4GLM-566.2#269
5GLM-4.763.8#380
6Kimi K2 (Thinking)63.1
7DeepSeek V3.2 ExpThinking55#439
8Qwen 3 235B A22B 2507 (Thinking)49.4#466
9Qwen 3.5 35B A3B48.1#492
10Kimi K247.2#443
11Gemma 3 27B (IT)43.8#1324
12Qwen 3 235B A22B 2507 Instruct39.4#616
13GPT-OSS-120BHigh35.8#704
14MiniMax-M2.530.6#600
15MiMo-V2-FlashNon-reasoning25#684

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

Interactive version: theaggregate.ai/benchmark?slug=lit-lost-in-translation-humanities-abstracts · How It Works · Data refreshed daily, snapshot 2026-11-21.