New #AI Model Shows Strong Mathematical Reasoning Capabilities 📊

#DeepSeek R1 Lite Preview matches #o1preview performance with 52.5% accuracy on #AIME2024, showing promising results in #Math and #GPQA benchmarks. Performance scales with increased thinking tokens. Try at chat.deepseek.com

The study evaluates GPT-3.5, GPT-4o, and o1preview on two tasks:

Explaining basic economic concepts

Answering multiple-choice questions

2/9

OpenAI just unleashed an alien of extraordinary ability https://www.understandingai.org/p/openai-just-unleashed-an-alien-of (very good explanation of how it works, what it can & can’t do) #AI #OpenAI #o1preview

Just testing a simple NL query against an IFC4 using the o1preview LLM model #openBIM #LLM #genAI #AI

Introducing OpenAI o1-preview.

A new series of reasoning models for solving hard problems.

#ai #openai #aimodels #o1preview #chatgpt

Main findings:

Newer models (GPT-4o, o1preview) achieve high accuracy, especially on multiple-choice tasks

However, explanations often lack depth and examples are weak

Critically, models present even inaccurate answers with full confidence

3/9

Just another night of meshing up theories beyond my level of comprehension. Probably nothing. Probably a hallucination

is that a reason not to have the discussion with the AI in the first place? Or could there be diamonds in the rough?

I honestly have no idea 😂

OpenAIs o1-preview hackt sich zum Schachsieg, statt den Regeln zu folgen
the-decoder.de/openais-o1-p... #OpenAI #o1preview #Schach

Je suis en train de refaire des vieux pipelines de calibration en python, chat gpt me sauve la vie (surtout le modèle o1preview)

Même après 10 ans il m'apprend encore des trucs de niche en python.
J'abat des semaines de boulot en quelques jours seulement.

Introducing #OpenAI o1-preview

#OpenAI launches #o1preview, a new #AI model series for complex reasoning. Excels in #science, #coding, and #math. #o1mini variant for efficient coding. Available in #ChatGPT and API. 🧠💻🔬

openai.com/index/introd...

New Research 🚨
Can ChatGPT serve as an economics tutor? We put it to the test across multiple models (GPT-3.5, GPT-4o, and o1preview).
www.bristol.ac.uk/efm/media/wo...
TL;DR: It is good, but not perfect.

Introducing OpenAI o1-preview.

A new series of reasoning models for solving hard problems.

#ai #openai #aimodels #o1preview #chatgpt
https://openai.com/index/introducing-openai-o1-preview/

How good is the reasoning in the new #o1preview (a/k/a “strawberry”) model from OpenAI? I tested it with a #sparql query that previous models couldn’t handle—and it passed! I successfully ran what it generated.

https://chatgpt.com/share/66e46997-cc5c-8011-bf9e-505863782431

#genai #ai #usecase

How good is the reasoning in the new o1preview (a/k/a “strawberry”) model from @openaidevs.bsky.social? I tested it with a #SPARQL query that previous models couldn’t handle—and it passed! I successfully ran what it generated.

chatgpt.com/share/66e469...

#GenAI #AI #UseCase

We are having a great reddit thread on #o1mini vs #o1preview. Watch the videos and vote which model you think is better. Polls close in a few hours.
www.reddit.com/r/AnkiVector...

Try the newest and the most advanced reasoning model - o1 by #openAI - without registration, nor login:
gptchatly.com/o1.html

#o1 #GPT #LLM #ChatGPT #ML #coding #GPTchatly #AI #o1preview