GPT 6.1 Terra lands 5th on ParseBench table parsing at ~1 cent per page
dotey · x · 2026-10-28
- LlamaIndex founder Jerry Liu calls GPT 6.1 Terra a strong frontier baseline for document parsing, with fresh results on ParseBench, LlamaIndex's open OCR/parsing benchmark.
- Strengths: spikes at table parsing, ranking 5th overall — ahead of 6 Sol, behind Opus 5.5 and Astra; decent content faithfulness and reading order.
- Weaknesses: struggles on layout, charts, and formatting.
- At 1 cent per page, it's a solid option for table-heavy pages, best paired with specialized OCR solutions like LlamaParse.
- Blogger dotey adds hands-on impressions: solid for translation and writing, faster and more durable than Sol; not used for coding.
- ParseBench evaluates five dimensions (tables, charts, faithfulness, semantic formatting, visual grounding) using human-verified pages and deterministic rules, with open dataset, code, paper, and leaderboard.
More from Models
- Letta's FrontierMemory leaderboard: Letta Dream tops at 73.3%, beating OpenAI and Claude agents — afurgs · 2026-10-28
- xAI Rebrands as SpaceXSI, Launches Grok 4.7 and Unified Multimodal API — RachelVT42 · 2026-10-28
- 'Benchmaxxed' Models Sink on Release as Community Waits for Next Attempts — rickasaurus · 2026-10-28
- Artificial Analysis launches Progress Index v1 to track capability since Nov 2022 — ArtificialAnlys · 2026-10-28
- AA v5 tasks consume 59k median output tokens, up two-thirds from v4.3 — ArtificialAnlys · 2026-10-28
- Effort settings now swing scores 20+ points at 10x cost, AA v5 finds — ArtificialAnlys · 2026-10-28