AI News
AI community posts, news and release updates
Latest updates
- News
ChatGPT Helps Discover Rare Gene Mutation, Brain Cancer Patient Gets Correct Diagnosis After 5 Years
Published Jul 24, 2026A patient diagnosed with glioblastoma and given months to live defied expectations for five years. Her husband used ChatGPT to interpret radiology reports, revealing an IDH1 mutation that led to a re-diagnosis of IDH-mutant astrocytoma with a better prognosis.
- ChatGPT
- 医疗AI
- 胶质母细胞瘤
- IDH1突变
- 真实故事
Opus 5 debuts at #2 on the Vals Index with 74.8%, less than a point behind Fable 5 (75.1%), and claims first on 14 of 27 benchmarks.
- benchmark
- Fable 5
- Opus 5
- Vals Index
- News
Claude Opus 5 now available in Cursor – matches Fable 5 performance at half price
Published Jul 24, 2026In Cursor, Claude Opus 5 scores similarly to Fable 5 on CursorBench (66.7 vs 66.5) at half the price and supports Zero Data Retention.
- Cursor
- 集成
- Claude Opus 5
- Idea
Opus 5 first look: breaks backward compatibility but shines when starting fresh
Published Jul 24, 2026After a week of testing, the author finds Opus 5 incompatible with existing workflows, stopping early or missing instructions. However, starting from scratch yields much better results. Medium or low thinking effort works better. The author calls it a 'poor man's Fable' and will likely use it only when Fable tokens run out.
- 测试
- 工作流
- 体验
- Opus 5
Reports indicate that Claude Opus 5 surpasses Fable 5 in multiple benchmarks while costing half as much.
- 对比
- Fable 5
- Opus 5
Anthropic introduces Opus 5, which exceeds Fable 5 on various benchmarks and is the new state of the art in many aspects, while remaining priced at Opus levels.
- Pricing
- Fable 5
- state-of-the-art
- Opus 5
Anthropic officially introduces Claude Opus 5, described as thoughtful and proactive, approaching Fable 5's frontier intelligence at half the price.
- 发布
- Claude Opus 5
- Idea
Opus 5 already beats Fable 5 in most benchmarks, and 30% on ARC AGI 3? wtf?
Published Jul 24, 2026A user is surprised that Opus 5 already beats Fable 5 on most benchmarks and scores 30% on ARC AGI 3, which is astonishing.
- 基准
- ARC-AGI
- Opus 5
- Idea
Most positive day for American AI - companies unite against regulatory capture
Published Jul 24, 2026Today I feel the most positive about the future of American AI. Companies coming together to fight regulatory capture is incredible.
- 监管
- 美国AI
- 行业联合
- Idea
Jacobian Conjecture Refutation Reveals Structural Limit of AI Interpretability
Published Jul 24, 2026The refutation of the Jacobian Conjecture suggests a fundamental mathematical limit to AI interpretability, potentially impacting future development of explainable AI.
- limits
- Jacobian conjecture
- AI interpretability