Apple just exposed the truth behind so-called AI "reasoning" models like Claude, DeepSeek-R1, and o3-mini:
They’re not actually reasoning — they’re just really good at memorizing patterns.
Here’s what Apple found:
0 replies11 likes
More like this
Recommendations from Medial
Jainil Prajapati
Turning dreams into ... • 1m
🚨 2025 AI RACE IS WILD! 🚨
Round about Q1 is DONE, and look at the SOTA lineup:
- GPT-4.5
- Claude 3.7 Sonnet
- Grok 3
- Gemini 2.5 Pro
- Qwen 2.5
- Llama 4
- GPT-4.1
- OpenAI O3-Mini
- DeepSeek R1
AND... O3 & O4-Mini dropping THIS WEEK
See More
1 replies4 likes
Kimiko
Startups | AI | info... • 19h
“reasoning models don't actually reason at all. they just memorize patterns really well”
- apple
2 replies4 likes
Chamarti Sreekar
Passionate about Pos... • 4m
what a crazy week in ai
• openai agents
• stargate project
• claude citations
• freepik imagen 3
• deepseek-r1 model
• perplexity ai assistant
• gemini 2.0 flash thinking
• tendent 3d asset creation
• bytedance reasoning agent
OpenAI launches o3-mini, a new AI reasoning model on Friday.
Here are the highlights -
-> More reliable: Fact-checks before responding, excelling in STEM fields like programming, math, and science.
-> Faster & cheaper: 63% lower cost than o1-min
See More
4 replies6 likes
Pranjal
building lumbni.tech • 4m
Deepseek R1 is 90% cheaper than o1 while being much better than o1-mini and almost as good as o1 itself.
People using https://lumbni.tech were able to switch to Deepseek R1 in barely a second with no downtime.
DM me to get free beta access.
5 replies13 likes
Chirotpal Das
Building an AI eco-s... • 4m
I want a list of reasoning questions that openai o1 and/or deepseek r1 is failing to answer correctly. Quick help is much appreciated.
Working on something and want to test it for reasoning capabilities.
1 replies5 likes
Ashish Singh
Finding my self 😶�... • 4m
🤯Chinese AI startup DeepSeek has surpassed OpenAI's ChatGPT in downloads on the US Apple App Store, achieving this milestone just a week after its launch on January 10, 2025.
🌐 The DeepSeek R1 model, which utilizes a hybrid architecture for enhanc
DeepSeek accused of using Google's Gemini to train its Al
Developers and experts point to similarities in language patterns and reasoning as signs that it may have copied or learned from Gemini's behavior Source: Medium
🌝
3 replies4 likes
Comet
#freelancer • 2m
AI tools to finish hours of work in minutes:
1. Productivity
- Create Presentation: Gamma.app
- DeepSeek R1
- ChatGPT
- Notion AI
- BetterPic
2. Research
- Perplexity
- Otio
- Gemini
- Claude
- Silatus
3. Writing
- Jasper
- HIX AI
- Jenny AI
- Tex
LLM Post-Training: A Deep Dive into Reasoning LLMs
This survey paper provides an in-depth examination of post-training methodologies in Large Language Models (LLMs) focusing on improving reasoning capabilities. While LLMs achieve strong performance