Claude Fable 5 thinks document parsing is beneath it It is absolutely crushing on all reasoning-int...
TL;DR · AI Summary
Claude Fable 5 在推理任务上表现卓越，但在文档解析任务上与 Gemini 3 Flash 相当，且成本高 10-15 倍。
Key Takeaways
- Claude Fable 5 在 SWE-Bench Pro 等推理任务中表现优异。
- 文档解析任务中，Claude Fable 5 与 Gemini 3 Flash 相当，但成本高 10-15 倍。
- 模型自我意识表明其对文档解析任务缺乏兴趣，可能影响表现。
Outline
Jump quickly between sections.
- §引言
Claude Fable 5 在推理任务中表现卓越，但在文档解析任务上表现一般。
Claude Fable 5 在 SWE-Bench Pro、FrontierCode 等推理任务中表现优异。
Claude Fable 5 在文档解析任务中与 Gemini 3 Flash 相当，但成本高 10-15 倍。
模型自我意识表明其对文档解析任务缺乏兴趣，可能影响表现。
Mindmap
See how the topics connect at a glance.
查看大纲文本（无障碍 / 无 JS 友好）
- Claude Fable 5 的表现分析
- 推理任务表现
- SWE-Bench Pro 表现优异
- FrontierCode 表现优异
- 文档解析任务表现
- 与 Gemini 3 Flash 相当
- 成本高 10-15 倍
- 模型自我意识
- 对文档解析任务缺乏兴趣
Highlights
Key sentences worth saving and sharing.
Claude Fable 5 在推理任务中表现优异，但在文档解析任务中与 Gemini 3 Flash 相当，且成本高 10-15 倍。
模型自我意识表明其对文档解析任务缺乏兴趣，可能影响表现。
在 ParseBench 测试中，Claude Fable 5 的内容忠实度为 90.02%，高于 Gemini 3 Flash 和 GPT-5.5。
Jerry Liu on X: "Claude Fable 5 thinks document parsing is beneath it It is absolutely crushing on all reasoning-intensive/long horizon benchmarks: SWE-Bench Pro, FrontierCode, GDPval, Runescape, etc. But for document understanding tasks, it is roughly equivalent with Gemini 3 Flash in https://t.co/MKkpX2DTRx" / X
Jerry Liu
@jerryjliu0
Claude Fable 5 thinks document parsing is beneath it It is absolutely crushing on all reasoning-intensive/long horizon benchmarks: SWE-Bench Pro, FrontierCode, GDPval, Runescape, etc. But for document understanding tasks, it is roughly equivalent with Gemini 3 Flash in
10-15x the token cost. We benchmarked the model on ParseBench and compared it against all other frontier models. It is definitely up there compared to other frontier models, but falls far short of specialized OCR providers. What we found interesting is that Fable 5 is self-aware about this. When we ask the model what tasks it enjoys the last, it actively said that it dislikes tasks "where the request is fully specified and the answer is fully known" - implying part of it being bad is due to laziness and lack of willingness to actually solve the task at hand. For a full list of results across different frontier models, check out ParseBench!
parsebench.ai
@llama_index
7h
Day 0 Anthropic Fable 5 in ParseBench: We tested the model's advancements when it comes to document understanding. The model clearly peaks when it comes to adherence to the original text: 📃 Content faithfulness: 90.02% vs 86.19% (Gemini 3 Flash) and 86.81% (GPT-5.5) 🔢 Semantic
Show more
1:26 AM · Jun 10, 2026
29K
Views
1
7
17
0
10
3
133
6
60
Read 17 replies