Meta's Llama 4 puts US back in lead to ‘win the AI race’ — David Sacks
The White House AI and crypto czar David Sacks says Meta’s release of its latest AI model, Llama 4, has pushed the United States into the lead in the global artificial intelligence race.“For the US to win the AI race, we...
The White House AI and crypto czar David Sacks says Meta’s release of its latest AI model, Llama 4, has pushed the United States into the lead in the global artificial intelligence race.
“For the US to win the AI race, we have to win in open source too, and Llama 4 puts us back in the lead,” Sacks said in an April 5 X post, as speculation continues to mount over the US and China competing for the top spot in the global AI race.
Sacks has been vocal about the AI race since taking on his role following US President Donald Trump’s inauguration on Jan. 20. Just over a week into the job, Sacks said he is “confident in the US, but we can’t be complacent.”
Llama 4 “best in their class for multimodality,” says MetaSack’s latest comment came after Meta’s AI division said in an X post on the same day that it is introducing the fourth generation of its Llama models, Llama 4 Scout and Llama 4 Maverick.
Source: David Sacks
“Our most advanced models yet and the best in their class for multimodality,” Meta said.
Meta said its Llama 4 Scout model has 17 billion active parameters and uses 16 experts.
The company claims it outperforms rival large language models — Gemma 3, Gemini 2.0 Flash-lite, and Mistral 3.1 — “across a broad range of widely accepted benchmarks.”
Meanwhile, Llama 4 Maverick also has 17 billion active parameters but is configured with 128 experts. Meta claimed the Maverick model can outperform GPT-4o and Gemini 2.0 Flash “across a broad range of widely accepted benchmarks.”
Llama 4 Maverick instruction-tuned benchmarks. Source: Meta
It also said Maverick can perform similarly to DeepSeek v3 on “reasoning and coding tasks” despite using only half the active parameters.
Related: NFT marketplace X2Y2 shuts down after 3 years, pivots to AI
Less than a year ago, in July 2024, Meta CEO Mark Zuckerberg said that in 2025, he expects Llama models to become “the most advanced in the industry.” It has been just over two years since Meta first released the limited version of Llama 1 in February 2023.
At the time, Meta said it was “blown away” by the demand, receiving over 100,000 requests for access.
Magazine: XRP win leaves Ripple a ‘bad actor’ with no crypto legal precedent set
Original source
Read on CointelegraphRelated market context
Anthropic’s dramatic model release strategy raises censorship risks, the shift to proprietary AI models is accelerating, and Chinese open source solutions are outperforming US counterparts | All-In Podcast
Chinese open source AI models surpass American counterparts, challenging global competitiveness and raising governance concerns. T...
Blockworks Acquires Messari in Deal Highlighting Crypto’s Data Consolidation Race
Bitcoin Magazine Blockworks Acquires Messari in Deal Highlighting Crypto’s Data Consolidation Race Blockworks, the New York-based...
Ripple chases AI’s machine economy as XRPL stablecoins near $1 billion
Stablecoin liquidity on the XRP Ledger (XRPL) has nearly doubled over the past month, putting the network within reach of a $1 bil...
Bitcoin Mining Cost Model Points To $47,000 Floor, But Analysts Urge Caution
TL;DR Crypto Rover says Bitcoin has never bottomed below electrical production cost, currently estimated at $47,000. Mining-cost m...
Metaplanet Acquires Siiibo Securities in Push to Build Bitcoin Financial Ecosystem
Bitcoin Magazine Metaplanet Acquires Siiibo Securities in Push to Build Bitcoin Financial Ecosystem Metaplanet Inc., Japan’s large...
MetaMask reports connectivity disruption across multiple blockchain networks
The disruption highlights the vulnerability of decentralized finance ecosystems to single points of failure, impacting user trust...