China’s AI Playbook: Baidu’s Models Cost 1% of OpenAI—What's Next?

Good morning AI entrepreneurs & enthusiasts,

China’s AI expansion is surging once again, with tech giant Baidu unveiling two advanced AI models at just 1% of the cost of OpenAI’s GPT-4.5 and half the price of DeepSeek’s R1.

With bold performance claims and dramatically reduced costs, we might be seeing the start of a global AI pricing battle.

In today’s AI news:

• China’s Baidu unveils ultra-affordable AI models
• Judge denies Musk’s attempt to block OpenAI’s restructuring
• Harvard researchers develop an AI agent for personalized medicine
• NVIDIA GTC 2025
• Top Tools & Quick News

China’s Baidu introduces ultra-affordable AI models

The News: Chinese tech giant Baidu has just launched two competitively priced multimodal AI models—ERNIE 4.5, a major upgrade to their foundational model, and a deep-reasoning model named Ernie X1.

The details:

• ERNIE 4.5 boasts superior “EQ” and language processing, as well as improved hallucination prevention, logical reasoning, and coding efficiency.
• Baidu claims ERNIE 4.5 surpasses GPT-4o across several benchmarks while costing only 1% as much—approximately $0.55 and $2.20 per million input and output tokens.
• ERNIE X1, Baidu’s inaugural reasoning model, rivals DeepSeek’s R1 at just half the price.
• Similar to DeepSeek’s R1, ERNIE X1 employs a structured, step-by-step “thinking” method, excelling in complex calculations and document analysis.

Why it matters: China is making AI accessible at an unprecedented scale. With ERNIE 4.5 priced at just 1% of GPT-4.5 and ERNIE X1 matching DeepSeek’s R1 for half the cost, the AI industry is on the brink of a significant price shakeup. Western competitors may be forced to reduce prices, expanding AI accessibility worldwide.

Judge denies Musk's attempt to halt OpenAI's restructuring

The News: A federal judge has rejected Elon Musk’s request for a preliminary injunction against OpenAI’s structural changes, while expediting the trial and dismissing several of Musk's claims, per a March 4, 2025 ruling., while expediting the trial and dismissing several of Musk's claims, per a March 4, 2025 ruling.

The details:

• The judge determined Musk failed to meet "the high burden" necessary for a preliminary injunction but allowed certain claims to proceed with a fast-tracked trial this fall.
• OpenAI cited internal communications suggesting Musk previously wanted to merge OpenAI into Tesla as a for-profit entity, contradicting his current stance.
• Musk sued OpenAI and CEO Sam Altman last year, arguing they deviated from their original mission of developing AI for humanity rather than corporate profits.
• OpenAI has denied these allegations, asserting that the non-profit remains intact and any structural shifts will ultimately enhance their mission.

Why it matters: With OpenAI potentially securing a $40B SoftBank investment to support its pivot to a for-profit model, this legal battle could significantly impact the company’s future and the evolution of its AI models. OpenAI’s valuation sits at $157B, making this one of the highest-stakes cases in AI today., this legal battle could significantly impact the company’s future and the evolution of its AI models. OpenAI’s valuation sits at $157B, making this one of the highest-stakes cases in AI today.

Harvard researchers introduce an AI agent for personalized medicine

The News: Scientists from Harvard and MIT have unveiled TxAgent, an AI system that uses multi-step reasoning and real-time biomedical knowledge retrieval to generate personalized treatment plans for patients.

The details:

• TxAgent employs 211 specialized tools to analyze drug interactions and contraindications in real time.
• It evaluates medications at molecular, pharmacokinetic, and clinical levels, factoring in age, genetics, and ongoing treatments.
• The AI synthesizes biomedical evidence and iteratively refines its recommendations through structured function calls.
• TxAgent’s toolkit, “ToolUniverse,” integrates with trusted data sources like openFDA and Open Targets.

Why it matters: TxAgent represents a leap forward in personalized medicine, offering AI-powered treatment recommendations to physicians. This advancement can streamline medical consultations, particularly in under-resourced regions.

NVIDIA GTC 2025

The News: NVIDIA's GPU Technology Conference (GTC) 2025 is set to take place from March 17 to March 21 in San Jose, California. This annual event is NVIDIA's largest conference of the year, focusing on artificial intelligence, accelerated computing, and related technologies.

The details:

• The conference is expected to draw approximately 25,000 in-person attendees and 300,000 virtual participants.
• CEO Jensen Huang will deliver the opening keynote on Tuesday, March 18, at 10 AM PT (1 PM ET) at the SAP Center. This address will be livestreamed for free on NVIDIA's website and YouTube channel.
• GTC 2025 will feature approximately 1,000 sessions with 2,000 speakers and nearly 400 exhibitors.

Why it matters: GTC 2025 promises to be a landmark event for developers, engineers, and tech enthusiasts interested in the cutting edge of AI and accelerated computing. With massive attendance and groundbreaking product launches, the event solidifies NVIDIA's role as a leader in AI hardware innovation.

Top Tools

• ?? OLMo 2 32B - First fully open model to surpass GPT-3.5 and GPT-4o mini
• ?? Freepik AI Video Upscaler - Enhance low-quality videos to 4K
• ?? Command A - Cohere’s ultra-fast enterprise AI with 256K context window
• ?? BrowserAgent - AI workflows in your browser with zero API costs

Quick News

• Google’s Gemini 2.0 Flash model is allegedly being used to remove watermarks from images.
• Figure launches BotQ, a factory aiming to produce 12,000 humanoids annually, scaling to 100,000.
• Patronus AI debuts a multimodal LLM-as-a-judge to enhance AI app reliability.
• Pika Labs unveils 16 new AI video effects.
• Sesame open-sources its realistic AI voice model CSM-1B.
• Vogent AI launches self-improving voice agents.
• Y Combinator CEO notes that 25% of their startups now use AI to write 95% of their code.
• OpenAI CPO Kevin Weil predicts 2025 will mark AI surpassing human-level coding abilities.