SpaceX issues $25 billion in bonds to restructure its debt and lower interest costs.

OKX will launch ARXUSD and UBUSD perpetual contracts. Trading for the ARXUSD perpetual contract will open at 14:30 Beijing Time on June 24, while the UBUSD perpetual contract will commence trading at 14:45 Beijing Time on the same day.

According to OnchainLens monitoring, whale address "0x913" has further added to its SOL short positions, currently holding a 20x leveraged short order of 624,680 SOL, with the position valued at approximately $43.4 million and a liquidation price of $82.84.

According to Blockaid's monitoring, the Yield Yak website has suffered a front-end attack. Subdomains of the site currently contain "eleven drainer" code. The attack method is similar to that of the earlier attack on Gitcoin.

According to data from Farside Investors, U.S. spot Bitcoin ETFs recorded a total net outflow of $113.8 million on June 23. BlackRock’s IBIT saw a net outflow of $182 million, while Fidelity’s FBTC, ARKB, HODL, and MSBT posted net inflows of $23 million, $31 million, $5.3 million, and $8.9 million respectively. For U.S. spot Ethereum ETFs, total net outflows reached $82.4 million. BlackRock’s ETHA had a net outflow of $86.1 million, ETHB saw a net outflow of $1.7 million, Fidelity’s FETH recorded a net inflow of $15.7 million, Grayscale’s ETH ETF posted a net outflow of $10.3 million, and the remaining products saw roughly flat fund flows.

Arthur Hayes’ family office Maelstrom has released a deep research report on Collector Crypt, a Solana ecosystem-based physical card trading platform, setting a target price of $4 for its CARDS token by the end of summer. According to GMGN data, CARDS is currently priced at $0.272, with a market capitalization of approximately $550 million. The on-chain tokenized Pokemon card market has seen explosive growth in recent months: gacha sales of the top seven platforms hit $230 million in May, up over 10x from $32 million in the same period last year. The core driver is the gacha mechanism: users spend funds to obtain random card-back NFTs, simulating the physical pack-opening experience while providing instant liquidity. Collector Crypt founder Tuom Holmberg previously revealed that gacha contributes 90% to 95% of the company’s business, and the platform has accumulated over $1 billion in total sales in 18 months since launch. Last year, the global physical card market reached $15.8 billion, and is projected to expand to $23.5 billion by 2030. Despite the fading of the crypto NFT boom, tokenization technology is reemerging as a key infrastructure for the card speculation market.

According to monitoring by Dongcha Beating, the Qwen team today officially launched Qwen-AgentWorld — the first native language world model that sets environment modeling as a training objective from the continued pre-training phase, rather than an after-the-fact adaptation of a general large language model. The model covers seven domains under a unified framework: text-based environments (MCP, Search, Terminal, SWE) and GUI-based environments (Web, OS, Android). Trained on over 10 million real-world interaction trajectories via three stages — CPT → SFT → RL — it enables cross-domain knowledge transfer. Also open-sourced simultaneously is the AgentWorldBench evaluation benchmark, with each test sample paired with observation data collected from real environment executions. Both the model and the benchmark are now available on Hugging Face and ModelScope. In the AgentWorldBench evaluation, Qwen-AgentWorld-397B-A17B achieved the highest overall simulation quality, outperforming GPT-5.4, Claude Opus 4.8, and Gemini 3.1 Pro. The research team also explored two application paths for world modeling in agent training: First, as a decoupled environment simulator, controllable simulated reinforcement learning (RL) can shape agent behavior and significantly outperforms RL trained exclusively in real environments. Second, as a unified agent base model, LWM warm-up training can be transferred to multi-turn agent tasks across seven benchmarks, three of which are not included in the training set at all, and no RL fine-tuning for agent tasks is required. The work preliminarily validates the potential of language world models as stronger foundational models for agents. Click the original link below to join the Dongcha Beating · Feishu AI News channel, which monitors global AI hotspots and news around the clock.