How to Access Free AI Models from OpenAI and Google

TL;DR

Free users can access OpenAI’s 03 Mini in ChatGPT by selecting Reason, or combine it with web search by selecting Search and Reason together. Google also offers Gemini 2.0 models free to developers and at lower API costs than competitors. OpenAI’s paid Deep Research adds extensive web-based analysis, so read on for access steps, performance comparisons, pricing, and limitations.

Transcript

well it's been another insanely busy week in the world of AI and I don't want to waste your time so let's get into this week's AI news breakdown starting with news that actually came out last week but I record these videos on Thursdays and this news came out on Friday of last week it was when open AI released their 03 mini now we did talk about it ... Read More

Key Insights

• OpenAI's 03 Mini model is now available to all users, including free tier users, making it one of the most accessible state-of-the-art AI models.
• Google's Gemini 2.0 models are released with significantly lower costs, making AI development more affordable for developers.
• The 03 Mini model excels in math and coding, outperforming many existing models except for the premium 01 Pro.
• Deep Research by OpenAI offers unparalleled research capabilities but is limited to Pro users at a $200 monthly fee.
• Google's Gemini models hold three of the top five spots in user preference rankings, showcasing their effectiveness.
• OpenAI's AMA revealed plans to increase context length and explore open-source strategies, indicating a shift towards more transparency.
• Chatbase is transforming customer service with AI agents capable of performing tasks beyond simple interactions.
• Google has removed its pledge against using AI for weapons and surveillance, indicating a shift in its ethical stance.

Questions & Answers

Q: How can free users access OpenAI’s 03 Mini model?

Free ChatGPT users can select the Reason button under the message composer to use 03 Mini. The model is available across every ChatGPT tier and through the API.

Q: Can free users combine OpenAI 03 Mini with web search?

Yes. Free users can select the Search and Reason buttons together to combine 03 Mini’s reasoning with search.

Q: How does OpenAI 03 Mini perform on math, science, and coding tasks?

The transcript says 03 Mini outperforms nearly every other listed model in math, except 01 Pro. The 03 Mini High version also trails only 01 Pro on PhD-level science questions, while performing well in coding and software engineering.

Q: What access do paid ChatGPT users receive for 03 Mini?

Pro users receive unlimited access to 03 Mini. Plus and Team users receive triple the rate limits provided for 01 Mini.

Q: Does OpenAI 03 Mini show its full chain of thought?

No. It displays a summarized version rather than the model’s complete reasoning process. The transcript notes that this makes incorrect answers harder to debug than systems that expose every reasoning step.

Q: What does OpenAI’s Deep Research do?

Deep Research searches the web and produces detailed analysis based on the user’s request and follow-up answers. In the example, it created an extensive YouTube strategy that was converted into a step-by-step checklist and a 4-week breakdown.

Q: How much does OpenAI Deep Research cost, and who can access it?

Deep Research is available to Pro users on the $200 a month plan. The speaker found it valuable but said the plan is probably not worth the cost for most people.

Q: What makes Google’s Gemini 2.0 models accessible to developers?

Google made its Gemini 2.0 models available free to developers and priced input at 10 cents per million tokens. The existing page compares that with GPT-40 at $10 per million tokens.

Summary & Key Takeaways

• Definition: OpenAI 03 Mini is a reasoning model available across every ChatGPT tier and through the API.

• Step 1: Free ChatGPT users select Reason under the message composer to access 03 Mini.

• Step 2: Select Search and Reason together to combine 03 Mini with web search.

• Who: Pro users receive unlimited 03 Mini access, while Plus and Team users receive triple the 01 Mini rate limits.

• When: Updated summarized chain-of-thought displays became available to free and paid users on February 6th.

• Compare: 03 Mini trails 01 Pro in the transcript’s math and PhD-level science comparisons.

• Number: OpenAI Deep Research is restricted to the $200 a month Pro plan.

• Number: Humanity’s Last Exam accuracy was 3.3% for GPT-40, 9.1% for 01, and 9.4% for DeepSeek R1.

• Number: The benchmark reported 13.0% accuracy for 03 Mini High and 26.6% for OpenAI with Deep Research.

• Tool: Deep Research searched the web and produced a YouTube strategy with a step-by-step checklist and 4-week breakdown.

• Number: Gemini 2.0 input costs 10 cents per million tokens, compared with $10 per million tokens for GPT-40.

• Who: Google offers Gemini 2.0 models free to developers, while its Gemini models occupy three of the top five user-preference rankings.