Happy Friday! Remember the rumored new AI model from OpenAI that was nicknamed Strawberry? Well, itās here, and itās a pretty huge update. But before I get to thatā¦
Here are the headlines:Ā
šļø Oprah hosted an AI special with Sam Altman and Bill Gates.
šŖ« Cohost, the would be X rival is running out of money and shutting down.
š The FDA has given its seal of approval to Appleās Hearing Aid feature.
OpenAI just released its first model with human-like reasoning abilities
OpenAI just unveiled o1, kicking off a new series of "reasoning" models designed to tackle complex questions faster than we can. They've also launched o1-mini, a smaller and more budget-friendly version. And yes, if you've been following AI buzz, this is the much-hyped Strawberry model everyone's been talking about.
This release is a big step toward OpenAI's goal of human-like artificial intelligence. Simply put, o1 outperforms previous models in writing code and solving tricky problems. But here's the catch: it's pricier and a bit slower than GPT-4o, so OpenAI is calling it a "preview" to highlight that it's still in the early days.
What's really interesting is how o1 was trained. Instead of just mimicking patterns from data, it uses reinforcement learningālearning through rewards and penaltiesāto figure things out on its own. It thinks in a "chain of thought," much like we do when solving problems step by step. OpenAI says it's more accurate and hallucinates less, though it's not perfect yet. This also makes for some great memes.Ā
The standout feature? o1 tackles complex tasks like coding and math while explaining its reasoning. Ideally, this should make it more accurate. While I was playing around with the model this morning, I did notice a lot fewer hallucinations. It's a promising leap forward in AI, even if it's still ironing out some wrinkles and comes with a heftier price tag.
Your agents stopped answering and started doing
Your agents call tools, write to databases and move money, so a wrong answer is now a wrong action, which is a considerably worse meeting. Arcjet runs policy before every LLM call, tool call, query and API request: prompt injection, sensitive data, bots, rate limits, spend quotas, and your own rules in Rego on top. It reads the OpenTelemetry you already emit, so switching it on is one environment variable, no code changes and nothing to deploy. Each check also reads the earlier steps in the same workflow, so an agent's tenth action gets judged against the first nine and not on its own. JavaScript, Python and Go SDKs all at 1.0, agent framework guards included, free tier to start on.
Daily Top Products
Products we love
EDUCATIONAL TOOLS
VELS is an AI voice-based learning and training simulations for mastering professional skills.
Fujiyama is an app that uses AI to make it easier to learn Japanese. You can discuss any topic with the AI assistant and save phrases for later.
MAKER TOOLS
JitBlox is an online visual environment for designing component-based web applications. You can use proven widgets from popular UI libraries to build your app.
Codemeup lets you keep your documentation code snippets up to date automatically by rendering snippets directly from your Github account.