OpenAI text-embedding-3-large
Supports 8,191 token context length with Matryoshka Representation Learning for flexible size reduction. Achieves 64.6% MTEB and 54.9% MIRACL benchmark scores, released January 2024.
Model Information
- Provider
- OpenAI
- License
- Proprietary
- Price per 1M tokens
- $0.130
- Dimensions
- 3072
- Release Date
- 2024-01-25
- Model Name
- text-embedding-3-large
- Total Evaluations
- 1917
Performance Record
Wins1068 (55.7%)
Losses571 (29.8%)
Ties278 (14.5%)
Wins
Losses
Ties
Performance Overview
ELO ratings by dataset
OpenAI text-embedding-3-large's ELO performance varies across different benchmark datasets, showing its strengths in specific domains.
OpenAI text-embedding-3-large - ELO by Dataset
Detailed Metrics
Dataset breakdown
Performance metrics across different benchmark datasets, including accuracy and latency percentiles.
FiQa
ELO 156269.2% WR166W-67L-7T
Accuracy Metrics
- nDCG@5
- 0.730
- nDCG@10
- 0.752
- Recall@5
- 0.700
- Recall@10
- 0.781
Latency Distribution
- Mean
- 10ms
- P50 (Median)
- 9ms
- P90
- 11ms
PG
ELO 155365.8% WR158W-76L-6T
Latency Distribution
- Mean
- 10ms
- P50 (Median)
- 10ms
- P90
- 11ms
SciFact
ELO 154362.5% WR150W-84L-6T
Accuracy Metrics
- nDCG@5
- 0.726
- nDCG@10
- 0.761
- Recall@5
- 0.768
- Recall@10
- 0.863
Latency Distribution
- Mean
- 11ms
- P50 (Median)
- 11ms
- P90
- 13ms
ARCD
ELO 154249.2% WR118W-58L-64T
Accuracy Metrics
- nDCG@5
- 0.899
- nDCG@10
- 0.899
- Recall@5
- 0.940
- Recall@10
- 0.940
Latency Distribution
- Mean
- 14ms
- P50 (Median)
- 14ms
- P90
- 16ms
business reports
ELO 153257.5% WR138W-88L-14T
Latency Distribution
- Mean
- 10ms
- P50 (Median)
- 10ms
- P90
- 12ms
DBPedia
ELO 152958.2% WR138W-82L-17T
Accuracy Metrics
- nDCG@5
- 0.648
- nDCG@10
- 0.641
- Recall@5
- 0.255
- Recall@10
- 0.377
Latency Distribution
- Mean
- 8ms
- P50 (Median)
- 8ms
- P90
- 9ms
MSMARCO
ELO 152949.2% WR118W-71L-51T
Accuracy Metrics
- nDCG@5
- 1.000
- nDCG@10
- 1.000
- Recall@5
- 0.123
- Recall@10
- 0.224
Latency Distribution
- Mean
- 8ms
- P50 (Median)
- 8ms
- P90
- 9ms
NorQuAD
ELO 152234.2% WR82W-45L-113T
Latency Distribution
- Mean
- 14ms
- P50 (Median)
- 14ms
- P90
- 16ms
Compare Models
See how it stacks up
Compare OpenAI text-embedding-3-large with other top embeddings to understand the differences in performance, accuracy, and latency.