Doubao-1.8-Pro: benchmark results
ByteDance's Doubao 1.8 Pro, the flagship closed tier of the Doubao 1.8 line launched at the December 2025 Volcano Engine FORCE conference with upgraded reasoning, multilingual, coding, and tool-calling capabilities. Provider: ByteDance. Released 2025-12-18. Access: API.
Aggregate Rating 1708 ± 50, rank #311 of 10916 rated models, from 27 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LOM Benchmark - CommonNeighbor | 100 | Accuracy (%) | 100 |
| LOM Benchmark - EntityRelation | 99 | Accuracy (%) | 100 |
| LOM Benchmark - TreeCheck | 100 | Accuracy (%) | 100 |
| LOM Benchmark - Neighbor | 98 | Accuracy (%) | 91.7 |
| TeleCom-Bench - Basic Theory | 70.14 | Macro-F1 (%; multiple-select telecom theory questions, mean | 85.7 |
| TeleCom-Bench - Core Network | 64.46 | LLM-judge score (%; open-ended core-network product-knowledg | 85.7 |
| TeleCom-Bench - Event Verification | 80.48 | Exact match (%; 4G load-imbalance checks over structured net | 85.7 |
| LOM Benchmark - BipartiteMatching | 83 | Accuracy (%) | 83.3 |
| LOM Benchmark - Predecessor | 99 | Accuracy (%) | 83.3 |
| LOM Benchmark - Description | 99 | Accuracy (%) | 75 |
| LOM Benchmark - Edge | 100 | Accuracy (%) | 75 |
| TeleCom-Bench - Root Cause Diagnosis | 61.33 | Exact match (%; mean of a fault-agent set and a single-domai | 71.4 |
Interactive version: theaggregate.ai/model?slug=doubao-1-8-pro · How It Works · Data refreshed daily, snapshot 2026-11-08.