Antioch, a U.S. startup focused on training and validating robots inside digital environments that closely mirror the real world, closed a $32 million Series A on Sept. 8 led by Greylock. Combined with an $8.5 million seed round completed earlier this year, the company has raised $40.5 million in a little over a year. PANews said Antioch is part of a broader wave of companies drawing fresh capital as Physical AI becomes a new investment focus across the AI sector. The report argues that the infrastructure once grouped under the metaverse narrative never really disappeared. Instead, its center of gravity has shifted. NVIDIA’s Omniverse, once described as a foundation for the metaverse, is now defined on NVIDIA’s website as a set of libraries and microservices for building Physical AI applications, with product focus moving toward industrial digital twins, robot simulation and autonomous driving. PANews also points to a growing stack of specialized companies rebuilding a practical version of the virtual world: NavVis in spatial data capture, NdotLight in SimReady 3D assets for robot simulators, and Treble in acoustic simulation. The common thread is not immersion for consumers, but realism for machines. In this version of the virtual world, robots use simulation as a training ground before entering the physical world.

Alibaba’s Tongyi Qianwen team has released Qwen-Image-2.1, a unified model for text-to-image generation and image editing. The visual generation component has 7 billion parameters, and a single checkpoint covers text-to-image creation, multi-reference image editing, localized editing, and transparent RGBA output. The model supports 2K resolution by default and includes native transparent image generation as well as multi-reference editing capabilities. Qwen-Image-2.1 is already supported in Diffusers, ComfyUI, and vLLM-Omni for research and evaluation. Commercial deployment is not open by default and requires a separate license application to the Qwen team. According to tests run by the team on its internal Qwen-Image-Bench benchmark, the model posted a composite score of 60.28, above the other listed open-weight models. The item cited MarkTechPost as the source.

Anchorage Digital has chosen LayerZero as the interoperability partner for its stablecoin issuance, according to a post from LayerZero on X. The arrangement will provide omnichain cross-chain support for stablecoins issued through Anchorage Digital Bank, N.A. LayerZero said institutional-grade stablecoins launched on Anchorage’s issuance platform will use its Omnichain Fungible Token, or OFT, standard for transfers across multiple networks. Those networks include Ethereum, EVM-compatible chains and Solana. LayerZero also said Tether’s USAT will be the first stablecoin from the bank to connect to LayerZero interoperability through the OFT standard. The disclosure came from LayerZero’s own social media post, and no additional details on timing or rollout terms were included in the source material.

Qwen has released Qwen-Image-2.1 and made the model weights public, combining text-to-image generation and image editing in a single 7B-parameter model. The release natively supports 2K images and transparent-background output, with editing positioned as the central upgrade in this version. According to the announcement cited by BlockBeats, the model can take up to 10 reference images at once, allowing users to recombine multiple people or products into one image. For localized edits, users can directly select areas, draw over them, or provide masks without switching to a separate editing model. Qwen-Image-2.1 can also generate RGBA images with an Alpha channel and extract subjects from photos. On the tooling side, Diffusers, ComfyUI, vLLM-Omni, and SGLang were already integrated on the first day. The release is not fully open for unrestricted use, however. While the weights are available, the model is distributed under the Qwen Research License, which allows only non-commercial research and evaluation. Commercial use requires separate authorization from Qwen.

Qwen has introduced Qwen3.8-Omni-Flash, its first multimodal model designed specifically for AI agents. According to Techub, the model can process audio and video at the same time and use tools on its own to handle tasks such as editing video blogs, translating video clips, and summarizing movie content. The report, citing The Decoder, said Qwen3.8-Omni-Flash performs nearly on par with Google’s Gemini 3.8 Flash in audio and video benchmark tests. Its API pricing, however, is only a fraction of the cost of Google’s model. The launch positions Qwen with a lower-cost offering in multimodal AI, centered on agent use cases that require both media understanding and tool use.

Jina AI has released jina-ocr-v1, a document parsing model that converts PDFs, scanned files, tables, and charts directly into Markdown. The company said the model was post-trained on top of DeepSeek-OCR and keeps its mixture-of-experts design, with roughly 3.4 billion total parameters and about 570 million parameters activated per token during decoding. Jina also added FastMTP speculative decoding. In Jina’s own tests, jina-ocr-v1 scored 91.14 on OmniDocBench v1.6, 0.89 points above DeepSeek-OCR-2. On olmOCR-Bench, it posted 83.4, which Jina said was 7.4 points higher than the DeepSeek-OCR base model it actually builds on. Throughput was a central part of the release. Under a single NVIDIA A100 with concurrency set at 32, Jina said the model reached 2.57 pages per second, the highest among 14 systems it tested, versus 2.10 pages per second for DeepSeek-OCR, or about 22% higher. The model weights are already available on Hugging Face under a CC BY-NC 4.0 license, and commercial use requires contacting Jina.

PANews published a weekly roundup covering a broad set of crypto and adjacent market stories, with a heavy focus on AI, regulation, tokenized equities, onchain distribution and fast-moving token narratives. The selection included first-person accounts from AI engineers who say automation is beginning to replace specialized work, commentary from investor Zhu Xiaohu on where startup moats may survive after AI tools become widely accessible, and a data-driven look at the 630 companies backed by Y Combinator this year. PANews also highlighted financing moves by major Chinese tech firms as they race to fund AI infrastructure. On the policy side, the digest pointed to the U.S. Securities and Exchange Commission’s five-year innovation exemption framework for tokenized NMS stocks, Hong Kong’s new digital asset roadmap, and the failed CLARITY Act vote, which PANews said ended 49-50. Market coverage ranged from Uniswap’s traffic gains tied to Arc and Robinhood Chain, to Arc mainnet’s first-day activity above 1 million transactions, and sharp moves in ARB, ZEC and NEAR. The roundup closed with a list of headline items spanning AI products, central bank policy, tokenized stock trading, security incidents and onchain fund flows.

a16z Crypto and Jito Labs argue that many financial institutions are treating permissioned networks as a prerequisite for compliance when current U.S. rules do not require that approach. In a new guide on permissionless networks and financial compliance, the firms say banks, broker-dealers and asset managers can build products and transact on open blockchain networks if they apply risk-based controls where they actually have operational control. The paper points to the Bank Secrecy Act, FinCEN enforcement guidance and sanctions rules as support for that position. The report says regulators do not expect institutions to eliminate every possible risk. Instead, they look for reasonably designed anti-money laundering and sanctions programs, with enforcement focused on systemic failures rather than isolated mistakes. It also cites the Office of the Comptroller of the Currency, which said in Interpretive Letter 1186 that banks may pay blockchain network fees and hold the crypto assets needed to do so. According to the authors, that undermines the idea that institutions must identify and screen every validator before using a permissionless network. The paper also argues that privacy is no longer a fatal obstacle for institutions on public blockchains, pointing to confidential transfer systems, provenance proofs and sanctions-related cryptographic proofs. It says recent U.S. legislation, including the GENIUS Act, follows the same logic by placing AML, CFT and sanctions controls at the application layer rather than on the underlying permissionless network itself.