WMT26 General Machine Translation (all languages) (WMT26)
We show this table for reference; we do not rank on it.
Machine-translation quality on the WMT26 general translation task, averaged across all evaluated languages and scored 0-100 by an LLM judge.
Benchmark score on WMT26 — October 29, 2026
We compile the WMT26 rows from provider self-reports. North Small Translate leads the table at 83.6%. We do not use these results to rank models overall.
1 modelMultilingualCurrentDisplay onlyUpdated October 29, 2026
Benchmark score table (1 model)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | North Small TranslateCohere | 83.6% | Not reported | Open |
About WMT26
Year
2026
Tasks
General-domain translation across 50+ languages (32 high-resource plus 18 additional)
Format
0-100 judged translation quality, averaged across languages
Difficulty
Professional machine translation
BenchLM currently stores provider-run WMT26 evaluations. Cohere's September 2026 run covers 50+ languages with GPT-5.6 Sol as the judge on a 0-100 scale, where 80-100 means perfect or minor errors. Because the judge, language set, and prompts are provider-chosen, rows stay display-only and are not directly comparable to official WMT human evaluations.
Freshness and provenance
Version
WMT26 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does WMT26 measure?
Machine-translation quality on the WMT26 general translation task, averaged across all evaluated languages and scored 0-100 by an LLM judge.
Which model scores highest on WMT26?
North Small Translate by Cohere currently leads with a score of 83.6% on WMT26.
How many models are evaluated on WMT26?
1 AI models have published results on WMT26 in the BenchLM catalog.
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 6,100+ subscribers.
One email each week. Unsubscribe anytime.