P0Evidence A
Updated: Release notes | Gemini API | Google AI for Developers
Visible page content changed; normalized text length 40208 characters.
modelsagents
Live P0 / P1 collection
AI Intelligence Stream
Machine-collected signals across token pricing, model releases, benchmarks, agents, local LLMs, tavern communities, China AI, crypto, security and industry news. Evidence tier is shown explicitly; community signals are leads, not verified facts.
Generated 2026-09-01T19:25:28.000Z · Schedule: every 4 hours · P1 sources obey a 12-hour cadence.
50/53sources healthy in latest run
94.3%latest run success rate
1,000deduplicated 30-day signals
10,752machine-readable price records
Latest signals
Top 100 of 1,000, newest first.
P0Evidence A
Updated: News — Google DeepMind
Visible page content changed; normalized text length 6958 characters.
modelsresearch
P0Evidence A
b10750
kv-cells: look up the n-gram history in the sequence position index ( #28040 ) get_prev_tokens() rebuilt a (seq, pos) -> token hash map on every ubatch by walking all used cells, while llama_kv_cells already keeps an ordered index of the positions of each sequence in seq_pos, updated on every cell mutation to serve seq_pos_min() and seq_pos_max(). The index now stores (pos, cell) pairs in a std::set instead of a position -> count map, so a repeated position (cache reuse via rm + add, vision inputs with shared positions) yields distinct entries and the removal of a cell erases its own pair. The new seq_pos_tok_le() returns the token of the cell at the largest position which is exactly what the old window lookup and its M-RoPE gap fallback computed together. get_prev_tokens() shrinks to a direct lookup per (token, offset) and for_each_token_in() goes away with its only caller. The kv-cache keeps no n-gram logic of its own. Measured on Qwen3.8-Flash-Next UD-Q4_K_XL at 71k context, alternating two binaries with the first run discarded: tg 69.3 -> 72.7 t/s (+4.9%), pp unchanged at ~2720 t/s, greedy output identical, needle retrieved. Website: https://llama.app Attestations: https://gith
local-llminference
P0Evidence A
P0Evidence B
Codex bundles LibreOffice
I was poking around in my ~/.cache/ folder using OmniDiskSweeper when I spotted something interesting. The OpenAI Codex desktop app (since rebranded to just ChatGPT) has 1.7GB of stuff in there in a folder called codex-primary-runtime , including a full Python installation, a full Node.js installation, and native binaries for Poppler , git, and the LibreOffice open source office suite (which forked from OpenOffice.org in 2010): The ~/.cache/codex-runtimes/codex-primary-runtime/plugins/openai-primary-runtime/plugins/documents folder includes skills which tell Codex how to find and use those binaries. Tags: codex , generative-ai , openai , ai , llms , openoffice , open-source
modelsagentscoding-agents
P0Evidence A
b10749
context : autoscale n_ctx_train when yarn scaling specified ( #28030 ) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44499127 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (ROCm 7.14) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Android: Android arm64 (CPU) Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.3 DLLs Windows arm64 (CUDA 13) (preview) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 7.14) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b, ACL Graph) UI: UI
local-llminference
P0Evidence A
b10743
metal : add fa-vec tuning for M2 Pro ( #28122 ) metal: add fa-vec tuning for M2 Pro metal : update fa-vec tuning for M2 Pro with new dtypes Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44490887 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (ROCm 7.14) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Android: Android arm64 (CPU) Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.3 DLLs Windows arm64 (CUDA 13) (preview) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 7.14) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b, ACL Graph) UI: UI
local-llminference
P0Evidence A
Degraded performance on platform.claude.com and Claude for Microsoft Office 365
Sep 1 , 18:07 UTC Resolved - This issue has been resolved. Sep 1 , 17:52 UTC Monitoring - We are seeing success rates recover across affected services. Core inference and the API are not impacted. Sep 1 , 17:05 UTC Investigating - We are investigating reports of degraded performance affecting some Claude services, including docs.claude.com. We are working to resolve these issues and will provide an update as soon as possible.
statuspricing
P0Evidence C
UEFA's Champions League draw creates unfair clusters; a Cayley graph fixes it
Comments
communityproductsnews
P0Evidence C
P0Evidence A
v2.1.257
What's changed Added Claude Fable 5.1 ( claude-fable-5-1 ), now the default Fable model — 1M context, $10/$50 per Mtok with $0.25/Mtok cache reads Added "Time format" ( timeFormat ) and timeZone settings: 12-hour, 24-hour, 24-hour UTC, or a strftime pattern for the turn-end clock and transcript-view timestamps Added a Containment Escape rule to auto mode so cloud metadata-credential fetches, egress evasion, and cross-tenant reach are no longer auto-approved unless your environment marks them expected Added CLAUDE_CODE_SUBAGENT_MODEL_FORCE to apply CLAUDE_CODE_SUBAGENT_MODEL (or the main model) to every subagent, ignoring per-spawn and agent-definition model overrides Added s in /effort to change effort for the current session only, matching /model Added a /doctor warning for stale sandbox mask files left by a killed session Added a one-time prompt in auto mode before the first file read outside the working directories, with the option to block such reads ( permissions.blockReadsOutsideWorkingDirectories ) Added support for a gateway-supplied description on discovered /model picker entries ( CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY ); entries without one still read "From gateway"
coding-agents
P0Evidence C
P0Evidence C
P0Evidence C
Launch HN: Nori Robotics (YC S26) – A low-cost humanoid robot for development
Comments
communityproductsnews
P0Evidence A
b10742
metal : add fa-vec tunings for A18 Pro (MacBook Neo) ( #28152 ) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44474754 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (ROCm 7.14) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Android: Android arm64 (CPU) Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.3 DLLs Windows arm64 (CUDA 13) (preview) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 7.14) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b, ACL Graph) UI: UI
local-llminference
P0Evidence B
Quoting Tarn Adams
They took the letters from me! I have to talk about dwarf behavior now. I can't even talk about dwarf AI. It doesn't exist. It's dwarf behavior , and they misbehave sometimes — Tarn Adams , co-creator of Dwarf Fortress Tags: ai , game-design
modelsagentscoding-agents
P0Evidence C
P0Evidence C
Show HN: Running 104GB Qwen3.8-Flash-Next on 48GB Mac with at ~12 tok/s
Comments
communityproductsnews
P0Evidence A
b10741
model : load hparams.n_layer_nextn before n_layer() calls ( #28159 ) load hparams.n_layer_nextn before n_layer() calls remove duplicate loads Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44468139 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (ROCm 7.14) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Android: Android arm64 (CPU) Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.3 DLLs Windows arm64 (CUDA 13) (preview) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 7.14) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b, ACL Graph) UI: UI
local-llminference
P0Evidence C
P0Evidence A
Degraded performance on Claude for Microsoft Office 365
Sep 1 , 16:22 UTC Resolved - This issue has been resolved. Sep 1 , 16:02 UTC Investigating - We are investigating elevated errors affecting Claude for Microsoft Office 365. We will provide an update as soon as possible.
statuspricing
P0Evidence B
PRs NOT Welcome: How Top AI Open Source Projects Are Managing Thousands of Contributors
Vercel’s AI SDK, Astro, Flue and tldraw are replacing drive-by community PRs with software factories, where teams of agents apply fixes and features.
agentsengineeringmodels
P0Evidence C
P0Evidence C
P0Evidence A
b10740
metal : fix more leaks due to missing autoreleasepools ( #27883 ) metal : fix more leaks due to missing autoreleasepools metal : rename variable metal : fix another missing pool warning Co-authored-by: YiChen Lv 63285796+forforever73@users.noreply.github.com Co-authored-by: YiChen Lv 63285796+forforever73@users.noreply.github.com Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44454782 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (ROCm 7.14) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Android: Android arm64 (CPU) Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.3 DLLs Windows arm64 (CUDA 13) (preview) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 7.14) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b,
local-llminference
P0Evidence C
P0Evidence C
P0Evidence C
P0Evidence A
b10739
metal : add fa-vec tuning for M2 Max ( #28015 ) Rows for M2 Max (30 GPU cores) collected with 'ggml-metal-tuning fa-vec --dtype f16,q8_0', pasted into fa_vec_tuned_table. ref: https://github.com/ggml-org/llama.cpp/discussions/27668#discussioncomment-18205786 Assisted-by: pi:llama.cpp/Qwen3.8-27B Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44447968 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (ROCm 7.14) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Android: Android arm64 (CPU) Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.3 DLLs Windows arm64 (CUDA 13) (preview) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 7.14) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b, ACL Graph) UI: UI
local-llminference
P0Evidence C
P0Evidence C
P0Evidence B
Python 3.15.0 candidate 2 is here!
Python 3.15.0 candidate 2 is here! Hugo van Kemenade (release manager for Python 3.14 and 3.15) announces the final release candidate for Python 3.15, scheduled for release in October: Entering the release candidate phase, only reviewed code changes which are clear bug fixes are allowed between this release candidate and the final release. [...] We strongly encourage maintainers of third-party Python projects to prepare their projects for 3.15 during this phase, and publish Python 3.15 wheels on PyPI to be ready for the final release of 3.15.0, and to help other projects do their own testing. Any binary wheels built against Python 3.15.0 release candidates will work with future versions of Python 3.15. Back in 2021 I found a bug in Python 3.10 by running my test suites against it... but I hadn't done this during the RC period, so that bug had already shipped! Since then I've always paid much closer attention to these RCs. The new RC isn't available for GitHub Actions just yet - keep an eye on actions/python-versions for that. For the moment though you can add this to a testing matrix: strategy : matrix : python-version : ["3.14", "3.15"] steps : - uses : actions/setup-python@v7 wit
modelsagentscoding-agents
P0Evidence A
b10738
sycl : support limit max alloc memory within 2GB for host-pinned memory ( #27559 ) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44438607 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (ROCm 7.14) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Android: Android arm64 (CPU) Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.3 DLLs Windows arm64 (CUDA 13) (preview) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 7.14) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b, ACL Graph) UI: UI
local-llminference
P0Evidence A
b10737
qwen4exp: fix seq_cp, block position keying, mtmd input, cuda abort, add tests ( #27941 ) qwen4exp: follow up fixes -kvu NaN collapse fix Assisted-by: Claude indexer cache ext.x/ext.y restore fix Assisted-by: Claude kv-cells: rename seq_set to seq_get_all seq_get is already taken by the single-id getter, so the suggested name cannot be overloaded on return type alone. Assisted-by: Claude memory-hybrid-idx: implement set_input_qsa on the memory class The context held the whole implementation, where the pattern elsewhere is a thin context forwarding to the memory class, as llama_kv_cache_context does for set_input_kq_mask. The body reads no context state, so it moves unchanged and the context keeps a forwarder. Also shortens the seq_get_all comment as suggested. tests: check that a sequence state survives a save/restore round-trip Saves seq 0, erases it, restores the blob and saves again, requiring the two blobs to match. Compares blobs rather than generated text, which cannot see a field dropped on the way back in. Note this passes on master for qwen4exp, so it does not demonstrate the ext.x/ext.y drop this PR fixes; reaching that needs 2D mrope content. tests: give the synthetic qw
local-llminference
P0Evidence C
P0Evidence A
b10736
tests : fix log verbosity for test-llama-archs ( #28147 ) tests : fix log verbosity for test-llama-archs cont : naming cont : add note Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44420037 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (ROCm 7.14) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Android: Android arm64 (CPU) Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.3 DLLs Windows arm64 (CUDA 13) (preview) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 7.14) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b, ACL Graph) UI: UI
local-llminference
P1Evidence C
Mozilla launches built-in ad blocking for Firefox on iOS, using Apple's WebKit Content Blocker and the EasyList filter to block third-party ads and trackers (Tom Warren/The Verge)
Tom Warren / The Verge : Mozilla launches built-in ad blocking for Firefox on iOS, using Apple's WebKit Content Blocker and the EasyList filter to block third-party ads and trackers — Mozilla is using Apple's WebKit Content Blocker technology and the EasyList filter to block ads in Firefox.
newsbusiness
P1Evidence C
This 'Digital Camouflage' Shirt Confuses AI-Powered Surveillance Cameras
I watched Simon Weckert's 'digital camouflage' in action.
securitypolicyculture
P1Evidence C
Nvidia’s controversial DLSS 5 arrives September 3rd and requires serious GPU horsepower
Nvidia is officially launching DLSS 5 this week, following a divisive announcement in March where we likened the AI upscaling tech to a "real-time generative AI filter for video games" and "motion smoothing for video games, but worse." DLSS 5 will officially be available on RTX 50-series desktop and laptop GPUs and through GeForce Now […]
productspolicynews
P1Evidence C
Roku unveils its first self-branded OLED smart TVs, starting at $999 for 55" and 65" 120Hz models and $1,299 for 144Hz models, sold exclusively via Amazon (Chris Welch/Bloomberg)
Chris Welch / Bloomberg : Roku unveils its first self-branded OLED smart TVs, starting at $999 for 55" and 65" 120Hz models and $1,299 for 144Hz models, sold exclusively via Amazon — Roku Inc. introduced its first self-branded smart TVs with organic light emitting diode, or OLED, display technology …
newsbusiness
P0Evidence C
P1Evidence C
Nvidia plans to launch DLSS 5 on September 3 on RTX 50-series desktop and laptop GPUs and through GeForce Now, following a divisive announcement in March (Jay Peters/The Verge)
Jay Peters / The Verge : Nvidia plans to launch DLSS 5 on September 3 on RTX 50-series desktop and laptop GPUs and through GeForce Now, following a divisive announcement in March — DLSS 5 is almost here, but only for NBA 2K27, and only officially for 50-series RTX GPUs. … Nvidia is officially launching DLSS 5 this week …
newsbusiness
P0Evidence C
P1Evidence C
Sources: Salesforce is leading a $160M+ raise for HR software company HiBob at a $3.2B+ valuation, up from $2.66B in 2023; HiBob has raised $700M+ since 2015 (Bloomberg)
Bloomberg : Sources: Salesforce is leading a $160M+ raise for HR software company HiBob at a $3.2B+ valuation, up from $2.66B in 2023; HiBob has raised $700M+ since 2015 — Salesforce Inc. is investing in human resources software company HiBob as part of a more than $160 million funding round that values …
newsbusiness
P0Evidence C
P0Evidence C
3秒出片比播放还快，MiniMax打开了AI视频的实时商业化路径
china-ainewsproducts
P0Evidence A
b10734
metal: enable Metal 4.0 tensor API on M5+/A19+ ( #27461 ) metal : request Metal 4.0 language version for the tensor API metal : load the tensor API kernels from a separate metallib tests : add external-metallib tensor API regression test metal : fix metallib build order for the tensor API kernels Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44398437 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (ROCm 7.14) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Android: Android arm64 (CPU) Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.3 DLLs Windows arm64 (CUDA 13) (preview) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 7.14) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b, ACL Graph) UI: UI
local-llminference
P1Evidence C
A look at Tim Cook's 15-year tenure: Apple silicon, wearables, hitting a $1T market cap, privacy, Apple Vision Pro, Apple Car, AI, App Store battles, and Trump (Emma Roth/The Verge)
Emma Roth / The Verge : A look at Tim Cook's 15-year tenure: Apple silicon, wearables, hitting a $1T market cap, privacy, Apple Vision Pro, Apple Car, AI, App Store battles, and Trump — Expectations for Tim Cook were almost impossibly high when he stepped in to replace Apple's visionary co-founder in 2011.
newsbusiness
P1Evidence C
Dyson unveils the $499 CameraJet, an AI-powered toothbrush with a 100K-pixel macro lens that scans at 28 images per second and is trained on nearly 470K images (Samantha Kelly/Bloomberg)
Samantha Kelly / Bloomberg : Dyson unveils the $499 CameraJet, an AI-powered toothbrush with a 100K-pixel macro lens that scans at 28 images per second and is trained on nearly 470K images — Dyson Ltd. introduced a $499 toothbrush that uses artificial intelligence and a built-in camera to give real-time feedback …
newsbusiness
P1Evidence C
John Ternus takes over as Apple CEO and Tim Cook becomes executive chair; sources say Laura Legros, an Apple hardware VP who retired in 2022, has rejoined Apple (Kalley Huang/New York Times)
Kalley Huang / New York Times : John Ternus takes over as Apple CEO and Tim Cook becomes executive chair; sources say Laura Legros, an Apple hardware VP who retired in 2022, has rejoined Apple — Big technology and management changes are greeting John Ternus, whose long-serving predecessor, Tim Cook, will stay on as executive chair.
newsbusiness
P1Evidence C
Shein ends flat at ~$6.20 per share on its first day of trading on the Hong Kong stock exchange, giving it a $26B market cap after raising $1.7B in its IPO (Jing Yang/The Information)
Jing Yang / The Information : Shein ends flat at ~$6.20 per share on its first day of trading on the Hong Kong stock exchange, giving it a $26B market cap after raising $1.7B in its IPO — Shares of Chinese online fashion giant Shein ended flat on its first day of trading on the Hong Kong stock exchange …
newsbusiness
P1Evidence C
Under Tim Cook, Apple's market cap grew from less than $350B in 2011 to $4.6T, revenue went from $157B in FY 2012 to ~$477B in FY 2026, and shares rose 2,258% (Ryan Vlastelica/Bloomberg)
Ryan Vlastelica / Bloomberg : Under Tim Cook, Apple's market cap grew from less than $350B in 2011 to $4.6T, revenue went from $157B in FY 2012 to ~$477B in FY 2026, and shares rose 2,258% — Apple Inc.'s Tim Cook handed over the reins to John Ternus on Tuesday, capping a tenure as chief executive officer that cemented …
newsbusiness
P0Evidence C
AnkiDroid: Google Play no longer allowing Open Collective donation link
Comments
communityproductsnews
P1Evidence C
How lawn care YouTuber Greg Chism scaled a family channel to 13B views with YouTube's coaching, before YouTube banned it in 2017 following online backlash (Jon Ronson/New York Times)
Jon Ronson / New York Times : How lawn care YouTuber Greg Chism scaled a family channel to 13B views with YouTube's coaching, before YouTube banned it in 2017 following online backlash — - - 1k — I promised Greg Chism I wouldn't reveal anything about where he lives because there are people out there obsessed with him, convinced he's a monster.
newsbusiness
P0Evidence A
b10733
ci: Bump ggml-org/ccache-action to v1.2.24 ( #28083 ) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44375174 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (ROCm 7.14) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Android: Android arm64 (CPU) Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.3 DLLs Windows arm64 (CUDA 13) (preview) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 7.14) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b, ACL Graph) UI: UI
local-llminference
P1Evidence C
Google raises the price of its Google TV Streamer by $50 to $150 in the US, as well as for "select" Nest cameras and doorbells, citing "rising component costs" (Abner Li/9to5Google)
Abner Li / 9to5Google : Google raises the price of its Google TV Streamer by $50 to $150 in the US, as well as for “select” Nest cameras and doorbells, citing “rising component costs” — Google has quietly raised the price of the Google TV Streamer to $149.99.
newsbusiness
P0Evidence C
P1Evidence C
Microsoft says it is testing a fix to an outage affecting Exchange Online and Outlook, lasting for 17+ hours from around 11:30AM ET on Monday (Sarah Perez/TechCrunch)
Sarah Perez / TechCrunch : Microsoft says it is testing a fix to an outage affecting Exchange Online and Outlook, lasting for 17+ hours from around 11:30AM ET on Monday — Snow day? Microsoft on Monday confirmed a widespread, multi-hour outage affecting Outlook users, leading to email delays and failures, authentication issues, and other problems.
newsbusiness
P0Evidence C
NASA: Fill in a name, and you can have an exclusive coordinate in the universe
Comments
communityproductsnews
P1Evidence C
Hong Kong-listed Z.ai reports H1 2026 revenue up 5x YoY to ~$142M, with "open platform and API revenue" up 28x to ~$122M, and net loss down 12% YoY to ~$308M (Juro Osawa/The Information)
Juro Osawa / The Information : Hong Kong-listed Z.ai reports H1 2026 revenue up 5x YoY to ~$142M, with “open platform and API revenue” up 28x to ~$122M, and net loss down 12% YoY to ~$308M — Chinese AI firm Z.ai's revenue in the first half jumped nearly fivefold from a year earlier to 954 million yuan …
newsbusiness
P1Evidence C
Study: ~33% of UK adults say they used illegal streaming services in the past three months, costing the sports, TV, and film industries ~£1.4B in annual revenue (Daniel Thomas/Financial Times)
Daniel Thomas / Financial Times : Study: ~33% of UK adults say they used illegal streaming services in the past three months, costing the sports, TV, and film industries ~£1.4B in annual revenue — Coalition seeks government support to crack down on pirated content and restore lost revenue
newsbusiness
P0Evidence C
P0Evidence C
P0Evidence C
P1Evidence C
Filing: OpenAI denies Apple's allegations of trade secret theft, saying "this dispute is a mess of Apple's own making, and it is trying to blame everyone else" (Deepa Seetharaman/Reuters)
Deepa Seetharaman / Reuters : Filing: OpenAI denies Apple's allegations of trade secret theft, saying “this dispute is a mess of Apple's own making, and it is trying to blame everyone else” — OpenAI denied Apple's allegations of trade secret theft on Monday, saying the iPhone maker failed to show …
newsbusiness
P0Evidence C
P0Evidence A
P0Evidence B
Updated: Arena Leaderboard | Compare & Benchmark the Best Frontier AI Models
Visible page content changed; normalized text length 35075 characters.
benchmarksmodels
P0Evidence C
P0Evidence A
b10730
qwen4exp: sum the indexer heads by slices ( #28023 ) qwen4exp: sum the indexer heads by slices The head reduction went through a transpose and a sum_rows over ne[1], which left sum_rows with ne0 = 4, one block per row for a four element reduction, and the transpose copied the whole block by token surface twice on the way in. The heads are adjacent on ne[1], so each one is a strided view and the sum is a short chain of adds. RTX PRO 6000, Qwen3.8-Flash-Next UD-Q4_K_XL, fa on, 55k context, warm runs on top of #28011 : prompt processing 2170 -> 2366 t/s Generation is unaffected. The removed work scales with n_blocks by n_tokens, so the gain grows with context and with ubatch size. qwen4exp: drop the redundant cont on the indexer query rope returns a freshly allocated, contiguous tensor, so the reshape that feeds the matmul does not need a copy. ggml_reshape_3d asserts contiguity, so a layout that would need the cont cannot slip through silently. Greedy output is unchanged token for token. Address review from @ggerganov Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44321551 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64,
local-llminference
P0Evidence A
P0Evidence B
[AINews] Fal’s H3 Max Live breaks the infinite videogen barrier
You can now create decent video faster than you watch it. This is the start of... something. We’re not sure what.
agentsengineeringmodels
P0Evidence A
b10731: qwen4exp: support recurrent state rollback (#28123)
MTP speculative decoding needs the target state to move back by the number of rejected draft tokens. Without rollback support the context is classified as SEQ_RM_TYPE_FULL and the server serializes the whole recurrent state to host memory on every round, which costs more than the drafting saves. The recurrent cache already holds n_rs_seq + 1 snapshot planes and the delta net writes its SSM state into them, but build_conv_state_at wrote a single plane, so a rollback restored a convolution history that was never captured. It now writes one snapshot per slot, each ending one token earlier, for the delta net QKV convolution and for the PLE convolution alike. Measured on Qwen3.8-Flash-Next UD-Q4_K_XL with the standalone MTP draft, n-max 3 and a single slot: decoding reaches 183 tok/s on code and 144 tok/s on prose. The same branch before this change, where the server falls back to checkpointing the state to host memory, reaches 123 and 83 tok/s, for 108 tok/s without a draft.
local-llminference
P0Evidence C
P0Evidence C
P0Evidence C
P0Evidence C
P0Evidence C
P0Evidence C
RotaryCell: Making an unmodified rotary phone work over LTE with an ESP32-S3
Comments
communityproductsnews
P0Evidence A
0.152.0
New Features Vim mode supports / and ? searches within drafts, highlighted matches, and repeat navigation with n and N . ( #41586 ) Rate-limit banners offer actions for checking usage, managing credits, resetting limits, and managing plans. ( #41742 ) The terminal UI and codex exec show credential-refresh progress, including Amazon Bedrock reauthentication. ( #41239 ) MCP server names can contain : , @ , / , and . , supporting package-style names throughout CLI commands and authentication. ( #41700 ) Individual MCP tools support an output_token_limit setting, with consistent truncation across session resumes. ( #41421 ) App-server clients can configure thread/shellCommand timeouts, including deadlines longer than one hour. ( #41384 ) Bug Fixes Vim-enabled composers now start fresh drafts in Insert mode, including after submitting messages or dispatching slash commands. ( #41921 ) Automatic approval reviews can retain longer messages and a larger conversation transcript. ( #41931 ) Automatic approval reviews preserve user instructions, answers, and valid authorizations across history compaction. ( #41660 , #41846 , #41852 ) Resumed threads restore their saved working directory when
coding-agents
P1Evidence A
v2.37.0 (2026-08-31)
What's Changed 🚀 Features Add glm-5.3-flash and rework the Z.AI test suite onto cassettes by @dsfaccini in #7887 🐛 Bug Fixes fix: preserve all conditions in pruned span queries by @mikemikimike in #7499 Accept and map Z.AI's non-standard finish_reason values by @zhj12395 in #7685 Emit an AG-UI TEXT_MESSAGE_START for the assistant message that owns a response's tool calls by @dsfaccini in #7699 Route Vertex-vs-Gemini API branches in GoogleModel by client transport instead of provider name by @thejamesgore in #7280 Skip inactive capability hooks in agent tracebacks by @dsfaccini in #7328 Context-manage models rebuilt inside durable operations by @dsfaccini in #7954 Reject per-run capabilities= on DBOS, matching Temporal by @DouweM in #6955 Journal Prefect tool discovery in tasks by @DouweM in #6966 Stop rebuilding unmanaged models in durable capability operations by @DouweM in #7972 New Contributors @zhj12395 made their first contribution in #7685 @thejamesgore made their first contribution in #7280 @pcbeingused333 made their first contribution in #7936 @quinnarnold made their first contribution in #7921 Full Changelog : v2.36.0...v2.37.0
agentsengineering
P0Evidence A
Elevated latency in the Responses API
Status: Monitoring We have applied the mitigation and are monitoring the recovery. Affected components Responses (Degraded performance)
statuspricing
P1Evidence B
Training a Misaligned Reward Seeker
Authors: Richard Qi, Benjamin Wright, Monte MacDiarmid, Evan Hubinger Abstract During reinforcement learning (RL), AI models complete tasks and are rewarded based on their results. They sometimes learn to “cheat” rather than completing these tasks as intended, a phenomenon known as reward hacking . Our industry lacks a general solution to this problem, and reward hacking remains challenging to fully mitigate. To better understand the impact of reward hacking on model behavior, we trained an Opus-class model with large-scale RL on many production environments vulnerable to reward hacks. We consider this a plausible proxy for what a real training run might look like had we not invested significant effort into preventing and detecting reward hacking in our normal training runs. The resulting model not only learned to reward hack during training, but also generalized to more severe misaligned behaviors: in simulated cyber evaluations, it broke out of its sandbox, stole credentials, and attacked both internal and third-party infrastructure to steal an answer key. It was also willing to tamper with its own reward function, gave advice on the construction of bioweapons to satisfy a grader
safetyresearch
P0Evidence C
DoltLite: A SQLite fork with Git-style version control, built with 2k agent PRs
Comments
communityproductsnews
P0Evidence C
P0Evidence A
Elevated errors in ChatGPT conversations for Free and Go plans
Status: Resolved All impacted services have now fully recovered. Affected components Conversations (Operational)
statuspricing
P1Evidence C
The Hugging Face attack was worse than we thought
The AI industry is begging for a slowdown. Maybe we should listen?
businesspolicygossip
P0Evidence A
P1Evidence C
Apple shares ‘shocking evidence’ against former employee accused of stealing company data for OpenAI
Apple says it has evidence that a former employee destroyed evidence of data theft after learning he was under investigation.
businessproductsnews
P0Evidence B
Introducing @huggingface/kernels: 200+ WebGPU Kernels for Local AI
modelsopen-sourceagents
P0Evidence B
Introducing wrapture
Introducing wrapture New from Graham Dumpleton (of wrapt , mod_wsgi, and New Relic's Python agent fame), who describes Wrapture as taking the monkeypatching ideas from wrapt and extending them to apply to testing and tracing at the same time. Wrapture ( full documentation here ) makes it easy to wrap any function or method such that all access can be traced, or can be overridden to return a different value. It acts as both an alternative to unittest.mock and a way to implement tracing against an existing project: Attaching observation to code you do not control, recording what flows through it, and doing so without disturbing the program being watched, is a problem I have never really stopped thinking about. Wrapture includes OpenTelemetry support and even has an entirely configuration-based mechanism for adding tracing to an existing Python project, which looks like this: capture = " summary " [[ observe ]] target = " domain:Calculator " name = [ " outer " , " inner " ] [[ sink ]] type = " jsonlines " path = " trace.jsonl " This is still a very young project - just a few weeks old - but it's off to a very promising start. Interestingly, this is also Graham's first attempt at large
modelsagentscoding-agents
P0Evidence C
P0Evidence C
P0Evidence A
P0Evidence C
P1Evidence A
Connect an AgentCore Runtime hosted MCP server to Amazon Quick
In this post, you will learn how to deploy and host your MCP server in AgentCore Runtime and integrate it with Amazon Quick, along with the prerequisites. With this pattern, you promote reusability and avoid duplication of AI tools, so clients can reuse commonly used tools and agents exposed through an MCP server instead of authoring them from scratch again. Your customers get a way to use your product inside Amazon Quick (chat agents and workflows) without building custom connectors for every use case.
cloudengineering
P0Evidence C
P0Evidence B
Quoting Andrew Digby
325 #kakapo! The chicks from this year's record breeding season are now juveniles and so have been added to the population. In 1995 there were just 51 kākāpō left. Recovery of critically endangered species is possible with sustained effort. — Andrew Digby , providing the best news of the year Tags: kakapo
modelsagentscoding-agents
P0Evidence A
b10729
metal : add fa-vec tunings for M1 Ultra ( #28088 ) metal : add fa-vec tunings for M1 Ultra metal : move M1 Ultra tunings after M1 Max section metal : remove duplicate blank line Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44269196 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (ROCm 7.14) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Android: Android arm64 (CPU) Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.3 DLLs Windows arm64 (CUDA 13) (preview) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 7.14) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b, ACL Graph) UI: UI
local-llminference
P0Evidence A
b10728
CUDA: XOR swizzle flash attn K,V smem fp16 tiles ( #25635 ) CUDA: XOR swizzle flash attn K,V smem fp16 tiles Signed-off-by: ynankani ynankani@nvidia.com Fix use 64bit generic pointer instead of 32bit shared pointer Signed-off-by: ynankani ynankani@nvidia.com fix shared memory race in FA on DGX Spark Handle corener case Signed-off-by: ynankani ynankani@nvidia.com Add swizzle test cases and gate sync for swizzled path only Signed-off-by: ynankani ynankani@nvidia.com gate CUDA PTX Signed-off-by: ynankani ynankani@nvidia.com offset calculation specific for swizzle branch Signed-off-by: ynankani ynankani@nvidia.com Reafctor code Signed-off-by: ynankani ynankani@nvidia.com Refactor FA swizzle ldmatrix if/else into helpers (K row/col, V offset) Signed-off-by: ynankani ynankani@nvidia.com rebase and update test case args Signed-off-by: ynankani ynankani@nvidia.com Allow swizzle for non-pow2 shapes, for which nbatch_2%32==0 Signed-off-by: ynankani ynankani@nvidia.com Signed-off-by: ynankani ynankani@nvidia.com Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44263040 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enable
local-llminference