Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Apple

AI overview

Sign in and the AI will write an overview from our coverage.

Headlines · 472

  1. Reolink Home Hub 2 launches at $199.99 with local AI event search, no subscription

    Reolink announced the Home Hub 2 at IFA 2026 and has put it on sale for $199.99. According to Android Authority, the home security hub stores footage from up to eight cameras locally and lets users search for key events using AI without a subscription fee; it also supports 2.5-inch SATA drives up to 16TB and keeps one microSD slot. The report notes that Google Home Premium and iCloud+ require paid subscriptions for their AI detection features such as faces, pets and vehicles.

    Android Authority · 🔥 0
  2. IDC reports 22% global smartphone shipment decline amid rising memory costs

    IDC reported that global smartphone shipments fell 22% year over year in Q3 2026, while Counterpoint and Omdia estimated declines of about 20%. IDC attributed part of the pressure to AI-related demand raising storage-chip prices, which squeezed vendor margins and led some manufacturers to raise prices, reduce specifications or discontinue entry-level models. TMTPost noted that channel shipments are not the same as consumer purchases, citing Counterpoint data that Chinese smartphone sell-through fell 13% during the four-week 618 shopping period.

    钛媒体 · 🔥 0
  3. Apple HomeView hub adds temp/humidity sensors, syncs with LG and Schneider thermostats

    IT之家, citing a MacRumors report, says Apple's first smart home hub, HomeView, has built-in temperature and humidity sensors that monitor indoor conditions and share readings with new LG Electronics and Schneider Electric thermostats, syncing with Apple Home Collection accessories. The report also says HomeView can use the iPhone's on-device AI to run an Adaptive Temperature feature that automatically adjusts air conditioning and ventilation based on occupancy and nighttime routines. According to Apple's notes cited in the report, the sensing is optimized for roughly 15–30°C and 30–70% relative humidity, accuracy may drop during prolonged loud audio playback, and the sensors need time to calibrate after startup.

    IT之家 AI · 🔥 0
  4. OpenAI agrees to avoid Apple-claimed trade secrets during lawsuit

    Bloomberg reports that OpenAI agreed not to use trade secrets that Apple alleges were stolen, for as long as their lawsuit plays out. OpenAI also agreed to collect forensic images of 14 employees' devices. Apple claims the secrets were taken by employees who defected from the iPhone maker, according to the report.

    Techmeme · 🔥 0
  5. TSMC posts sixth straight record profit on AI demand, guides Q4 higher

    TSMC reported third-quarter 2026 net profit of NT$763.1bn, up about 69% from a year earlier, on revenue of NT$1.49tn, up roughly 51% — both above estimates and a sixth consecutive quarterly profit record. The company guided fourth-quarter revenue to $52.3bn–$54.3bn and raised its 2026 dollar revenue growth outlook to about 44%; high-performance computing was 68% of revenue, smartphones 21%, and 3nm chips 33% of wafer revenue. Chairman C.C. Wei said AI demand keeps rising and the supply-demand gap remains large, with agentic AI driving CPU and networking chip demand, while striking a cautious note on consumer and price-sensitive markets for 2027; TSMC kept 2026 capital spending at $60bn–$64bn and said it will add another 3nm fab in Tainan, with its second Arizona fab starting mass production in Q3 2027, ahead of schedule.

    IT之家 AI · 🔥 0
  6. Apple and CMU's SSR turns agent reasoning into multiple choice, cutting latency over 90%

    Researchers from Apple Machine Learning Research, Carnegie Mellon University and collaborators introduced Selection-based Structured Reasoning (SSR), a framework that replaces freeform per-step reasoning in multimodal search agents with selection from pre-specified, reusable natural-language candidates, using the model's own likelihoods and no auxiliary task head. The authors report that on seven multimodal search benchmarks with 2B and 4B models built on Qwen3-VL, SSR matched same-scale leading search agents in average success rate while cutting per-turn reasoning latency by over 90% and total per-question inference latency by 28–54%. Machine Heart (机器之心) further reports a 61.37% average success rate for 4B SSR with GRPO and 51.26% for the 2B version; all figures are author-reported with no independent replication supplied.

    Apple Machine Learning Research · 🔥 0
  7. Meta releases Muse Spark Open, an 816B open-weight multimodal model

    Meta released the weights of Muse Spark Open on 15 October, an open-weight mixture-of-experts model with about 816 billion total and 42 billion active parameters and a context window of up to one million tokens. It accepts text, images, audio and video as input and produces text, and Meta trained it for agentic coding, tool use and long-horizon tasks. Meta, which calls it the most capable US open-weight model to date, worked with Apple so the model can run locally on two M5 Ultra Mac Studio desktops via MLX; it is also offered through third-party APIs including Together AI, Baseten, Fireworks AI, Groq, Modal, Nebius and OpenRouter, with weights on Hugging Face under the Muse Model License. Arena.ai ranked it the top US open model on its Code Arena: WebDev leaderboard at 1511 points and 13th among open models overall, while an Artificial Analysis score of 39 cited in community discussion placed it behind GLM 5.3 and Kimi K3.

    Hugging Face · 🔥 0
  8. Apple MLX runs 816B Muse Spark Open locally on Mac Studio cluster

    Apple Machine Learning Research published a post on October 15, 2026, detailing how the latest MLX release runs the nearly trillion-parameter Muse Spark Open (816B total, 42B active) locally on Mac Studio. It says a single Mac Studio offers up to 512GB of unified memory and can scale across multiple Macs using an RDMA over Thunderbolt driver; on a cluster of four M5 Ultra Mac Studios, the model generated over 36 tokens per second at an 8,192-token context. Apple also compared M5 Ultra with M3 Ultra and benchmarked Mistral Medium 3.5, Qwen 3.8, and pipeline and tensor parallelization schemes, claiming up to 4.6x the start-up time-to-first-token speed of the previous Mac Studio generation.

    Apple Machine Learning Research · 🔥 0
  9. Apple study compares plan-first vs retrieve-first tool discovery for LLM agents

    Apple Machine Learning Research published an EMNLP paper by Mingfei Zhang and co-authors comparing two ways tool-using LLM agents find APIs before planning: Retrieve-then-Plan (RtP), which embeds the query and retrieves similar tools, and Plan-then-Retrieve (PtR), in which the model first generates abstract tool stubs that become retrieval queries. On ToolBench's 46,984 API endpoints across 50 categories, RtP beat PtR on single-tool queries while PtR beat RtP on multi-tool queries; the crossover held at retrieval budgets k=1–20, registry sizes from 1k to 47k APIs, and with Gemini 2.5 Flash and gpt-oss-120b. The authors also describe a stub-hallucination failure mode, note that PtR's extra discovery call roughly doubles inference cost, and suggest complexity-adaptive routing using RtP's top-1 retrieval score.

    Apple Machine Learning Research · 🔥 0
  10. Apple HomeView hands-on: an ambient smart display built around Apple's ecosystem

    TechRadar's Jacob Krol spent more hands-on time with HomeView, Apple's first smart display, in Apple's demo space. He describes an interface that feels like a supersized, customizable Apple Watch face you swipe between, with core apps such as Home, Music, FaceTime and Calendar preinstalled; developers cannot build dedicated HomeView apps, so Apple instead brings compatible experiences over from the iPhone. The piece notes a Dynamic Island-style music control strip when switching tasks, no notifications at all by design, iPhone Handoff for music, calls and Safari pages via a tap on the display, and a 12-megapixel camera that also serves as a mirror. Krol stresses this was brief demo-space time, not a full review or an at-home test, and the article also mentions refreshed HomePod mini and Apple TV 4K models that are said to bring Siri AI to more devices around the home.

    TechRadar · 🔥 0
  11. OpenAI says ex-employee took 100+ files on unreleased device to Apple

    In a filing made in federal court in San Jose, California, on Wednesday, OpenAI alleged that a former employee who spent about six months at the company before returning to Apple transferred more than 100 confidential files about at least one unreleased OpenAI device to personal accounts, including device images, engineering and manufacturing data and supplier information; some files were reportedly sent on June 2, his last day. OpenAI also said the person presented its product plans to senior Apple leaders, including Apple's vice president of operations in Asia, and that it is "very concerned" its trade secrets are now circulating inside Apple. Apple told Reuters it has policies preventing new employees from using a prior employer's confidential information and found no evidence supporting the allegations; per Apple's court filing published by 9to5Mac, Apple told OpenAI on October 12 that it could not locate any OpenAI confidential information and asked for specifics, and OpenAI filed with the court instead of replying.

    Gizmodo · 🔥 0
  12. iOS 27.2 Beta 4 adds five languages to Siri AI

    Apple released the iOS 27.2 Beta 4 developer preview (build 24B5110a) on October 14, 2026, adding French, Japanese, Korean, Portuguese and Spanish to Siri AI, according to IT之家. With the existing English support, Siri AI now covers six languages after the update; IT之家 said Apple had previously confirmed these language expansions, while availability still depends on the user's region and device. IT之家 also described Siri AI as Apple's newest AI assistant, capable of understanding personal context, performing app actions, sensing the screen, understanding images and drawing on world knowledge, with a dedicated app offering visual intelligence and writing/editing tools. DoNews published a similar report the same day but noted that its article was automatically generated by an open model.

    IT之家 AI · 🔥 0
  13. Apple HomeView vs Amazon Echo Show 11: which smart display fits your home

    TechRepublic compared Apple's HomeView and Amazon's Echo Show 11 across price, display, AI assistant, smart-home support and privacy. HomeView, unveiled October 13 at $299, pairs a 7-inch screen and A18 chip with Siri AI and facial recognition for up to six household members, and ships November 13 in eight markets including the US and UK but not the European Union. Echo Show 11 has been available since November 2025 at $219.99 with an 11-inch screen, Alexa+ and Zigbee hub support, while wall-mounting HomeView adds a mount starting at $59.95, and the report says HomeView's reliability and audio still need independent testing after November 13.

    TechRepublic · 🔥 0
  14. Apple and Duke's LensVLM browses compressed page images, then zooms in to read

    Machine Heart reports that Apple's foundation model team and Duke University proposed LensVLM, which first views compressed page images that hold the full text, then calls a tool to expand and read the pages relevant to a question, so the model can still consult the source after compression. The paper reports 68.9% average accuracy at 4.3x effective compression across seven text QA benchmarks, close to the 72.4% full-text reference, while at 15x initial compression the base model read compressed images directly at 18.1% versus 52.1% for LensVLM. Built on Qwen3.5-9B-Base and trained with supervised fine-tuning plus DAPO reinforcement learning for page selection and answering, the work was accepted at NeurIPS 2026 and a public LensVLM 9B demo is online.

    机器之心 · 🔥 0
  15. Apple Intelligence, Qwen added to China iPhone filings; December launch planned

    Multiple China-market iPhones received updated regulatory filings on September 29 carrying three large-model registration numbers: Apple Intelligence (Shanghai-AppleZhiNeng-202506160057), Tongyi Wanxiang (Shanghai-TongYiWanXiang-202410090024) and Tongyi Qianwen (ZheJiang-TongYiQianWen-20230901), according to IT之家 and cnBeta. The filings cover iPhone 15 Pro / Pro Max and later models, plus iPhone Air and the foldable iPhone Duo — matching the device list that supports Apple Intelligence. Apple Technology Development (Shanghai)'s Apple Intelligence model was registered on July 8, 2026 for use on Apple phones, and Alibaba has said Qwen will be integrated into Apple Intelligence to bring text and image understanding and content generation to Chinese users across iOS, iPadOS, macOS and visionOS. Bloomberg reports Apple plans to launch Apple Intelligence in mainland China in December via iOS 27.3, without Siri AI features.

    IT之家 AI · 🔥 0
  16. Apple site alt text hints full-size HomePod may get Siri AI

    IT之家, citing X leaker @itspdfu, reported that code and image descriptions on Apple's website suggest the full-size HomePod may support Siri AI in the future; IT之家 found that Apple's China site now describes a product image as showing a "multi-colored Siri AI glow on top." Apple has separately confirmed that HomePod mini 2 will support Siri AI, but not at launch — the capability is due in a "future software update," and requests involving personal context require pairing with an Apple Intelligence-enabled iPhone or iPad. 9to5Mac cautioned that Apple has since replaced the images, so the clues could be a website error or an unannounced plan. cnBeta, DoNews and AIbase repeated the same clues, while CNET reported that older HomePod minis will not get the more advanced Siri AI features and that some older Apple TVs will.

    IT之家 AI · 🔥 0
  17. Microsoft closes at record $543.45 as AI confidence lifts value by $1.4T since June

    Microsoft shares rose 0.4% to close at $543.45 on Tuesday, the first record close in nearly a year, surpassing a previous peak set in October 2025, IT之家 reported citing Bloomberg. The stock has gained more than 54% from its June low, adding about $1.4 trillion in market value, which the report ties to growing investor confidence in Microsoft and its position in AI. TechNews said Microsoft's market value returned above $4 trillion and cited a foreign brokerage target of $600, while Bloomberg-tracked analysts had an average target of $584.

    IT之家 AI · 🔥 0
  18. Apple researchers unveil first tri-modal masked diffusion models trained from scratch

    Apple Machine Learning Research published a NeurIPS paper introducing what the authors describe as the first tri-modal Masked Diffusion Models pretrained from scratch on text, image-text and audio-text data. The study systematically analyzes multimodal scaling laws, modality mixing ratios, noise schedules and batch-size effects, and proposes an SDE-based reparameterization that decouples physical batch size from logical batch size so the optimal batch size no longer has to be tuned. The authors also pretrain a preliminary 3B-parameter model on 6.4T tokens, reporting strong results in text generation, text-to-image and text-to-speech tasks; these are self-reported and not independently measured, and the paper's claim to be the largest systematic open study of multimodal discrete diffusion is author-asserted.

    Apple Machine Learning Research · 🔥 0
  19. Apple researchers detect code-switching directly from speech encoders

    A paper from Apple Machine Learning Research proposes a lightweight method that detects code-switching — alternating languages within one utterance — directly from speech, without relying on accurate multilingual transcription. The approach uses frame-level language information in pretrained speech encoders: it identifies language-selective hidden-state dimensions from monolingual recordings and aggregates their frame-level activations into a code-switching prediction. With a small code-switching set used to tune a decision rule, the authors report the Whisper-small encoder reaches a mean F1 of 0.798 across 14 language pairs in CS-FLEURS, versus 0.776 for a CNN and 0.751 for a HuBERT linear baseline trained for monolingual language identification.

    Apple Machine Learning Research · 🔥 0
  20. Apple ships macOS 27.2 developer beta 4, adding five Siri AI languages

    Apple released macOS 27.2 developer preview Beta 4 to Mac users on October 14, carrying build number 26B5112a and arriving seven days after the previous Beta/RC, according to IT之家. The report says Apple has not publicly listed specific fixes for Beta 4, noting that builds this close to an RC typically focus on bug fixes, performance and stability. macOS 27.2's headline change is described as Siri AI gaining support for French, Japanese, Korean, Portuguese and Spanish.

    IT之家 AI · 🔥 0
  21. OpenAI vows not to use Apple secrets as judge allows limited expedited discovery

    At a hearing in San Jose, OpenAI agreed not to use trade secrets Apple alleges were taken by employees who left the iPhone maker, and to collect forensic images of 14 employee devices, an agreement enforceable by US District Judge Edward Davila. Davila earlier granted Apple limited expedited discovery into OpenAI's interview materials for former Apple candidates and some LINE messages, and approved depositions of Chang Liu, Alyssa Peng and OpenAI chief hardware officer Tang Yew Tan, while denying expedited access to R&D areas such as metal finishes and power management. Hours before the hearing, OpenAI filed claims that a former employee who joined OpenAI and later returned to Apple sent more than 100 confidential files about an unreleased device to a personal account and presented plans to Apple leadership; Apple said it found no evidence supporting the allegation and accused OpenAI of deflection.

    9to5Mac · 🔥 0
  22. Bloomberg: Meta releases new AI model it calls 'frontier capabilities'

    Bloomberg Technology's Ed Ludlow reported that Meta released a new artificial intelligence model that the company says has "frontier capabilities" as it competes with OpenAI and Anthropic. The segment did not name the model or disclose its size, release date or any benchmark results, and the "frontier capabilities" description is Meta's own characterization. The same broadcast also covered Apple's new smart home lineup including the HomeView Hub and a chip selloff spreading from Asia to the US.

    Bloomberg Technology · 🔥 0
  23. New Mac app RemoveMacAI deletes Apple Intelligence, freeing space

    TechRadar reviewed RemoveMacAI, a Mac app that removes Apple Intelligence from macOS. The tool first gained traction in late September as a Terminal command; the same developer has now shipped an app wrapping that command, with a built-in tutorial and per-feature toggles for Siri AI, the ChatGPT extension, Writing Tools, Image Playground and more. In the reviewer's test, deleting Apple Intelligence freed 11.7GB, in line with the developer's claim of up to roughly 12GB, but the app is unsigned, so macOS refused to run it without manual approval, and it also requires installing a custom profile; undoing the changes reverts settings and re-downloads the models when needed.

    TechRadar · 🔥 0
  24. Apple launches fourth-gen Apple TV 4K with Siri AI and A19 chip

    Apple announced the fourth-generation Apple TV 4K on October 13; it ships November 13 at $299 in the US and HK$2,399 in Hong Kong. The box uses an A19 chip that Apple says delivers 50% faster CPU and up to twice the GPU performance than the previous model, adds three far-field microphones for hands-free Siri AI powered by Apple Intelligence, and includes 128GB of storage, Wi-Fi 7, Bluetooth 6 and a Siri Remote with Find My support. TechRadar reported that Apple confirmed Siri AI will also reach second- and third-generation Apple TV 4K models through a free software update in select countries and territories, with timing not yet announced.

    IT之家 AI · 🔥 0
  25. Apple launches HomePod mini 2 with future Siri AI

    Apple announced the HomePod mini 2 with an S10 chip, five colors, artwork-responsive “Sound Glow” lighting, and upgraded connectivity. It can serve as a dedicated Mac speaker and join second-generation HomePods for surround sound with Apple TV 4K, while Siri AI is promised through a future free software update in selected markets rather than at launch. The US price is $149.99, and Apple says the first-generation HomePod mini will not support Siri AI because of its older chip and memory configuration.

    TechRadar · 🔥 0
  26. Dongshan Precision clears HKEX hearing for $2–3B Hong Kong listing

    Dongshan Precision passed its Hong Kong Stock Exchange listing hearing on October 11 and has begun pre-marketing a main-board offering expected to raise about US$2 billion to US$3 billion, with UBS, Guotai Haitong, GF Securities and CITIC Securities as joint sponsors, according to a CLS report republished by TMTPost. The report says the company describes itself as the world's largest PCB maker for edge AI devices and an Apple supply-chain member; its roughly 6 billion yuan purchase of Source Photonics in 2025 added optical modules, and Source Photonics accounted for 16.02% of consolidated revenue and 52.92% of profit in the first quarter of 2026. First-half 2026 revenue was 27.798 billion yuan, up 64%, with net profit of about 2.985 billion yuan, up 293.5%, while its A-share market value has fallen from more than 500 billion yuan in June to about 240 billion yuan.

    钛媒体 · 🔥 0
  27. IMF warns AI share correction could threaten global financial stability

    The IMF’s latest Global Financial Stability Report warns that high AI valuations, rising debt and circular financing could amplify a market correction, naming an AI asset-price correction, higher government borrowing costs and a major cyber-attack as three salient near-term risks. It reviewed 14 publicly disclosed deals since 2023 in which hyperscalers and other investors put about $300bn of equity and $650bn of debt into financially weaker AI firms in exchange for $1.4tn in future revenue commitments; hyperscalers now account for about 10% of investment-grade bond issuance, up from under 1% in 2024. The IMF’s bank stress test found lenders broadly resilient but said banks holding 22% of global banking assets would be weak, and it sketched a hypothetical scenario in which global GDP, investment and consumption fall 0.4%, 1.3% and 0.6% by 2028; at the Bangkok annual meetings, Managing Director Kristalina Georgieva said AI is a key growth driver but over-reliance on one engine carries risks.

    Hacker News · AI(100+ 分) · 🔥 0
  28. Samsung's first AI glasses leak: ~41g, Snapdragon AR1, November launch

    IT之家 reported on October 13, citing SammyGuru, that Samsung's first smart glasses are expected to go on sale in November, weighing about 41 grams with standard lenses and carrying a Snapdragon AR1 Gen1 chip, 2GB of RAM, 64GB of storage, a 12MP camera, two open speakers and a four-microphone array; battery life is rated at up to 9 hours for the glasses and up to 72 hours with the charging case, with IPX4 water resistance. The glasses have no display, so AI voice replies play through the speakers; frames come from collaborations with Gentle Monster and Warby Parker, with the Gentle Monster versions sold in Korea and the US and the Warby Parker versions US-only. AIbase reported that the glasses run Android XR with deep Gemini integration and may be called Galaxy Glasses, but Samsung has not announced a price or exact release date, and the specs have not been officially confirmed.

    IT之家 AI · 🔥 0
  29. Apple study: bilingual data mixing beats hyperparameter tuning in data-scarce pre-training

    Apple Machine Learning Research published a paper, "Mix, Don't Tune," comparing two remedies for data-constrained pre-training: aggressive hyperparameter tuning such as high weight decay, and mixing in data from a high-resource auxiliary language. Across roughly 1,000 pre-training runs at four model scales from 150M to 1.43B parameters, using Arabic as the low-resource target and English as the auxiliary, mixing improved validation loss and downstream task accuracy more than hyperparameter tuning, with the gap growing with model size. The authors report gains equivalent to 2–3× the unique target data on validation loss and 2–13× on downstream accuracy, and argue that target-language validation loss systematically underestimates the value of mixing.

    Apple Machine Learning Research · 🔥 0
  30. Apple Creator Studio update adds AI writing in Freeform, Final Cut Pro Cinematic editing

    Apple announced a Sept. 29 Creator Studio update: Final Cut Pro and Motion on Mac can now add Cinematic effects to video captured on iPhone 18 Pro, letting editors adjust focus points and depth of field on the timeline, while Compressor can convert the newer Cinematic effects into the previous Cinematic format and adds HEVC and MPEG-4 export presets for social sharing. Pixelmator Pro for iPad gains the Select and Mask tools previously available on Mac for tricky edges such as hair and fur, and Freeform adds folder organization, collaborator invites, automatic Dark Mode plus Apple Intelligence-powered Writing Tools that can rewrite, proofread and summarize handwritten notes. Apple also says Creator Studio subscriptions require a personal Apple Account and cannot be enrolled with Managed Apple Accounts, and that Keynote, Pages, Numbers and Freeform remain usable without a subscription though premium and AI features require one, with device and OS requirements varying by app.

    TechRepublic · 🔥 0

← PreviousNext →

Experience and discussion from the community

Share my Apple experienceAsk about Apple

cua: Scale computer-use 2.0 with open-source drivers, cross-OS sandboxes, and benchmarks for training, evaluation, and data generation.

What problem it solves Computer-use projects need environments where agents can operate applications, run code, and interact with graphical interfaces within the same task. They also need repeatable ways to create tasks, evaluate agents, export trajectories, and generate data for training and research. Key features Shared desktop Spaces — Cua Spaces gives agents full desktops hosted on computers o…

github.comhttps://github.com/trycua/cua

agent-device: Open source device layer for AI coding agents.

What problem it solves Coding agents need a consistent way to inspect, control, debug, and verify applications across mobile, web, TV, and desktop targets. They also need repeatable workflows and reviewable evidence across local devices, simulators, CI, and remote device clouds. Key features Inspect and control app interfaces — Agents can read accessibility snapshots, act through refs and selector…

github.comhttps://github.com/callstack/agent-device

HermesEnabler: Enable Hermès Watch Faces on any Apple Watch devices

What problem it solves Hermès Watch Faces require an enabling process on supported Apple Watch devices. Users need instructions for installing or building the app, as well as guidance for preserving data and restoring original settings after updates or removal. Key features Install the IPA — The README provides a release-based installation flow using the **Watch** app after sideloading the IPA wit…

github.comhttps://github.com/pookjw/HermesEnabler

llama.cpp: Plain C/C++ implementation for local and cloud LLM and VLM inference.

What problem it solves Running large language and vision-language models can require substantial setup and may depend on the available hardware. The project aims to make inference accessible across a wide range of local and cloud hardware with minimal setup. Key features Minimal-setup inference — The project provides installation options for common platforms, including scripts, Docker, pre-built b…

github.comhttps://github.com/ggml-org/llama.cpp

TensorFold: A native language-model inference engine for serving checkpoints on Apple Silicon and NVIDIA GPUs.

What problem it solves Serving language models requires reading checkpoints, tokenizing requests, and running accelerated kernels while managing a local model server. TensorFold addresses these requirements with a native binary that can run on Apple Silicon and NVIDIA GPUs without requiring Python or MLX for serving. Key features Native Metal and CUDA inference — The Zig-based engine executes mode…

github.comhttps://github.com/ashhart/TensorFold

iloader: A user-friendly sideloader for installing apps and managing pairing files on iDevices.

What problem it solves Installing sideloaded apps and importing pairing files can require confusing device setup, certificate handling, and troubleshooting steps. The project provides a desktop workflow for installing IPA files, managing pairing files, and addressing common setup errors. Key features App installation — Install SideStore, optionally with LiveContainer, or import any IPA file. Sourc…

github.comhttps://github.com/nab138/iloader

iSponsorBlockTV: A self-hosted YouTube for TV client that automatically skips sponsor segments and handles advertisements.

What problem it solves Sponsor segments, introductions, and advertisements interrupt videos on YouTube for TV clients. A self-hosted application can automate this handling while supporting both network discovery and manual device setup. Key features Automatic segment skipping — The application connects to a YouTube for TV app and automatically skips segments such as sponsors and introductions. Sou…

github.comhttps://github.com/dmunozv04/iSponsorBlockTV

sglang-omni: A multi-stage serving runtime for omni, speech, and TTS models.

What problem it solves Interactive multimodal and speech systems often require generation to span heterogeneous stages with different compute patterns, dependency structures, and resource needs. Serving these pipelines requires coordinated stage lifecycles, inter-stage transport, model integrations, and a unified API surface. Key features Coordinated multi-stage runtime — Models generation as coor…

github.comhttps://github.com/sgl-project/sglang-omni

ipatool: A command-line tool for searching Apple App Store apps and downloading app packages.

What problem it solves Users need a command-line workflow to search for apps across Apple platforms and obtain .ipa or macOS .pkg packages from the App Store. The workflow also needs to support App Store authentication, purchases, version metadata, and automation. Key features Search across Apple platforms — Search the App Store for iOS, iPadOS, tvOS, watchOS, visionOS, and macOS applications. Sou…

github.comhttps://github.com/majd/ipatool

opencode: An open source AI coding agent that runs in the terminal, on the desktop, and via package managers.

What problem it solves Developers need a coding agent that can assist with development work while remaining inspectable and open source. The project addresses the need for configurable agents with different access levels, a straightforward installation path across platforms, and a way to explore codebases without modifying them. Key features Two built-in agents with different access levels — A def…

github.comhttps://github.com/anomalyco/opencode

uzu: A high-performance inference engine for AI models.

What problem it solves Developers need to run AI models inside applications while handling model downloads, inference configuration, and chat sessions. The project presents local inference with data privacy and no inference costs as its deployment model. Key features High-level application API — The engine exposes a simple API for creating an engine, loading a model, opening a chat session, and re…

github.comhttps://github.com/trymirai/uzu

AnyPS5: A tool for automatic porting of PS5 executables to Linux and Windows.

What problem it solves PS5 executables need to be converted into native executables for Linux and Windows, with suitable system libraries available for dynamic linking. The project addresses this porting work without emulation or a separate runtime process. Key features Native executable relinking — The relinker converts executables into the native format of the target operating system and provide…

github.comhttps://github.com/boykopovar/AnyPS5

timesfm: A pretrained time-series foundation model for time-series forecasting.

What problem it solves Time-series forecasting requires models that can forecast univariate and multivariate series across different context lengths. Forecasters also need to incorporate optional covariates and produce quantile forecasts without requiring per-task tuning. Key features Univariate and multivariate forecasting — Forecast individual time series or multiple target variates from NumPy a…

github.comhttps://github.com/google-research/timesfm

LocalAI: An open-source AI engine for running models locally across modalities and hardware.

What problem it solves Running AI models locally can require integrating multiple inference engines, model formats, hardware accelerators, and API surfaces. LocalAI addresses this with a small core, on-demand backends, and compatible APIs that work across supported hardware without requiring a GPU. Key features Composable inference backends — Backends are packaged separately and pulled only when a…

github.comhttps://github.com/mudler/LocalAI