AI & Web3 News 7/24/2025

Author

Stylianos Kampakis
July 24, 2025

AI News

  1. GPT-4 Turbo previewed with 128K context window: At OpenAI DevDay, OpenAI unveiled GPT-4 Turbo with a 128K-token context window, enabling it to process the equivalent of over 300 pages of text in a single prompt and offering improved performance at a reduced cost per token OpenAI.

  2. Google localises Gemini 2.5 Flash for Indian developers: At its Bengaluru I/O Connect India event, Google announced local processing support for Gemini 2.5 Flash in data centres in Delhi and Mumbai—addressing data residency and latency issues—and integrated Gemini capabilities into Firebase and AI Studio to accelerate secure, low-latency app development The Times of IndiaThe Economic Times.

  3. Anthropic’s Claude set to gain memory and chat search features: Anthropic is testing memory and chat-search functionality for its Claude AI assistant, allowing it to recall previous interactions and personalize follow-up responses, with these enhancements currently in trials for Claude’s mobile apps Lifewire.

  4. Anthropic upgrades Claude with shareable AI apps and experiences: Anthropic rolled out Artifacts, a feature in Claude enabling users to generate standalone AI-powered apps, documents, and tools without coding, which can be shared and iterated on collaboratively Anthropic.

  5. European Commission issues AI Act compliance guidelines for systemic-risk models: On July 18, 2025, the European Commission released guidelines to help providers of AI models considered to pose systemic risks comply with the upcoming EU AI Act, detailing requirements for risk assessments, incident reporting, and transparency measures ahead of the August 2, 2025 enforcement date Reuters.

Web3 News

  1. Ethereum’s Dencun upgrade introduces proto-danksharding: The Ethereum network activated the Dencun upgrade, which implements proto-danksharding (EIP-4844) to reduce Layer 2 rollup data-posting costs by over 90%, cutting gas fees and improving scalability for developers Consensys.

  2. Uniswap v4 goes live across 12 chains with modular pools: Uniswap v4 was deployed on Ethereum, Polygon, Arbitrum, Base, BNB Chain, Avalanche, and seven additional networks, introducing on-chain hooks for custom liquidity pools that enhance capital efficiency and allow developers to embed bespoke DeFi logic uniswapfoundation.mirror.xyz.

  3. Solana network boosts block capacity by 20%: On July 22, 2025, Solana increased its per-block compute limit by 20% to 60 million compute units, aiming to process more transactions per block and alleviate congestion CoinGape.

  4. Solana proposes 66% block size increase through SIMD-0286: Developers submitted SIMD-0286, a proposal to raise Solana’s per-block compute cap from 60 million to 100 million units (a 66% increase), to accommodate growing demand and future-proof the network’s capacity CoinDesk.

  5. Square rolls out Bitcoin payments for merchants using Lightning Network: Block’s Square has begun rolling out Bitcoin payments for sellers through the Lightning Network, enabling near-real-time settlement and automatic fiat conversion, with full availability expected across its seller base by 2026 CoinDesk.