Open source models 100 times cheaper than GPT: the revolution of accessible AI
How much does it cost to use AI in your company? The answer changed this week. Researchers from Neon published an important finding: they can replicate the accuracy of GPT-5.6 Sol (one of the most advanced models) by spending 100 times less money using open source models.
Think of it as the difference between hiring a Google engineer versus a talented programmer who taught himself. Both can solve the problem, but one costs much less.
Why does this matter? Until now, accessing powerful AI meant relying on companies like OpenAI, Google, or Anthropic, paying prices that favor large corporations. A small business could spend thousands of dollars monthly on ChatGPT queries. With these new models, that cost could be reduced to tens or hundreds of dollars.
Open source models are not free, but they are democratizing. You can install them on your own servers, customize them for your industry, and have full control over your data. There's no dependency on external vendors.
This opens opportunities especially for companies in Latin America that operate on tight margins. A digital agency using AI to automate reports, a logistics company that optimizes routes, or a call center that implements chatbots, can now do so without breaking the bank.
The cost of AI is collapsing. Whoever takes advantage of it first wins.
Source: Hacker News
What does this mean for you?
If you've rejected implementing AI because it's "too expensive," it's time to reconsider. Open source models now offer quality comparable to GPT at a fraction of the cost. Consult with a developer about using open source alternatives in your business.