Text Generation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
24.05TPOT (ms/token)
4
May 8, 2026
0.71MAUVE
4
Feb 26, 2026
0.329ROUGE-L
4
Feb 26, 2026
72.5Throughput (tokens/sec)
4
Feb 26, 2026
50.5Throughput (token/s)
4
Feb 26, 2026
63.6Coherence Win Rate
4
Feb 26, 2026
77.1Coherence: CD Preferred
4
Feb 26, 2026
88.7Coherence: CD Better Rate
4
Feb 26, 2026
17.2BLEU
4
Feb 26, 2026
23.2BLEU
4
Feb 26, 2026
2,520Average End-to-End Latency (ms)
3
Jun 23, 2026
3Coherent Emission Rate
3
May 15, 2026
34.2S-BLEU
3
Mar 4, 2026
80.94F1 Score
3
Mar 4, 2026
47.9BLEU-1
3
Feb 26, 2026
48.1BLEU-1
3
Feb 26, 2026
0.974Distinct-1
3
Feb 26, 2026
72.51Prefill Latency (ms)
3
Feb 26, 2026
62.3Throughput (token/s)
3
Feb 26, 2026
5.5Latency (s)
3
Feb 26, 2026
91.4BLEU-2
3
Feb 26, 2026
51.4Fluency Win Rate
2
Jun 12, 2026
3.56Speedup
2
Jun 2, 2026
10.69BLEU
2
Apr 21, 2026
1.45BPC
2
Feb 26, 2026