Search: “Seed 2.0 Pro”

Smart search ranks by meaning as well as keywords (one row per story, last 45 days).

[AINews] DeepSeek v4.1-Flash: 763B-P8B-D16B novel causal Encoder–Decoder architecture with vision marks the Return of the Whale

DeepSeek released V4.1-Flash, an open-weight 763B-parameter model with a novel causal encoder-decoder architecture, 1M context, vision input, and MIT license.

DeepSeek launched V4.1-Flash, an open-weight MIT-licensed model using a novel causal encoder-decoder architecture with 763B total parameters and asymmetric active parameters: 8B for prefill and 16B for decode. It supports 1M-token context and text+image input, priced at $0.30 per 1M input and $1.20 per 1M output tokens with a 50% off-peak discount. Artificial Analysis scored it 40 on its Intelligence Index, above DeepSeek V4 Pro 0813, and Vals ranked it the #1 open-weight model ahead of Kimi K3. Baseten shipped day-0 support and Ollama began rolling it out to paid subscribers.

Latent Space·Sep 12, 05:56 UTC · 4d agoModel release 7 sources1

80

visionOS 27.0 RC (24M362)

Apple seeded visionOS 27.0 release candidate build 24M362 to developers ahead of the general release.

Apple released the visionOS 27.0 release candidate (build 24M362) through its developer releases feed. The listing only provides downloads and release notes, with no security content or CVE details. RC builds typically precede the public availability of the final OS version.

Apple software releases·updated Sep 14, 17:00 UTC · 1d ago·first Sep 9, 17:00 UTC · 6d agoAdvisory 2 sources2

12

Can LLMs Engineer Their Own Agent Harness? ByteDance Seed’s HarnessDev Says Only 34 of 64 Changes Generalize

ByteDance Seed's HarnessDev benchmark finds LLM-built agent harnesses trail human engineering on code and search, with only 34 of 64 revisions generalizing.

Researchers from ByteDance Seed, SUTD, Georgia Tech, M-A-P, and TokenWave.AI introduce HarnessDev, a benchmark that evaluates the runnable agent harness an LLM writes rather than its answers, using Creation and Evolution stages across SWE-bench Pro, Terminal-Bench 2.1, MLE-bench, EQ-Bench3, and BrowseComp (2,207 instances). Six creator models including Opus 4.8, GPT-5.5, Gemini 3.1 Pro, DeepSeek V4 Pro, Qwen 3.7 Max, and Seed 2.0 Pro were tested; Opus 4.8 posted the best average of 67.8 versus an 86.2 human-engineered reference. Self-built harnesses beat references on writing and ML experimentation but lag badly on code and search, and quality proved executor-specific: Opus 4.8 fell from 69.3 to 33.0 on SWE-bench Pro when the executor was switched to Gemini. Evolution gains were small and noisy: of 64 adjacent changes, feedback and held-out scores agreed only 34 times (53.1%), and much generated state and memory code never executed.

MarkTechPost·Sep 11, 22:01 UTC · 4d agoAI research 2 sources

48

[AINews] Hot Chips: OpenAI’s Jalapeño, Cerebras CS-5, Groq 3 LPX, Apple M6

OpenAI unveiled Jalapeno custom inference chip claiming 1.5-1.9x better perf-per-watt than NVIDIA GB200/GB300, deploying in-house by year-end.

At the 37th Hot Chips conference, OpenAI published first benchmark details for its custom Jalapeno inference chip, claiming 1.5-1.9x more work per watt, 1.7-3.6x lower end-to-end latency, and 2.1-4.1x higher interactive-workload performance versus NVIDIA GB200/GB300, with the 700W-rated part staying at or below 550W in tests. Deployment into OpenAI's own infrastructure begins by year-end, with Gen 2 deep in development and Gen 3 underway. OpenAI also said GPT-Astra and Codex helped write low-level kernels, reportedly 1.5-1.8x faster than human-expert code for selected attention and MoE blocks. Cerebras CS-5, Groq 3 LPX and Apple M6 were also featured at the conference.

Latent Space·Aug 27, 01:31 UTC · 20d agoAI industry

72

DeepSeek launching v4.1 flash cheaper and more capable than v4 pro

DeepSeek will launch V4.1 Flash around September 10, 2026, claiming it outperforms V4 Pro on performance, cost, and speed while routing Pro traffic to it.

DeepSeek announced V4.1 Flash, launching around September 10, 2026, stating it surpasses V4 Pro across performance, cost, speed, and task completion time. After launch and before V4.1 Pro ships, all Pro requests will be routed to Flash and billed at Flash pricing. Off-peak prices are $0.003 for input cache hits, $0.15 for cache misses, and $0.60 for output, with peak-hour rates doubled.

Hacker News · security·Sep 9, 11:19 UTC · 7d agoModel releaseHN 44↑ · 5 comments

55

ukisai/Swift-Qwen3.8-27B-GGUF — new model trending #30 on Hugging Face

UkisAI released Swift-Qwen3.8-27B GGUF, a Qwen3.8-27B derivative cutting thinking tokens by 58.3% with under 1% performance loss and roughly 1.95x speedup.

UkisAI released Swift-Qwen3.8-27B as GGUF on Hugging Face, a reasoning-efficient derivative of Qwen3.8-27B using a Swift adapter that reduces median thinking tokens by up to 58.3% while keeping performance losses under 1% and delivering a 1.95x speed-up on several tasks. Reported benchmarks include GPQA-Diamond 88.28%, MMLU-Pro 84.95%, C-Eval 90.62%, AIME 2026 94.00% and Terminal-Bench 2.1 65.84%. The model is trending at #30 on Hugging Face, with BF16 weights and enterprise licensing also available.

Hugging Face trending models·Sep 11, 14:58 UTC · 4d agoModel release

42

⚡ Weekly Recap: VMware Exploits, Windows 0-Day, MCP Attacks, Browser Hijacks and More

Weekly recap: suspected China-nexus APT exploited VMware vCenter CVE-2026-59310, Lazarus used Windows zero-day CVE-2026-68820, and macOS flaw dropped Monero miners.

The week's top stories included a suspected China-nexus APT exploiting VMware vCenter CVE-2026-59310 (CVSS 9.8), a directory-traversal flaw enabling arbitrary code execution, with backdoors, reverse SSH and Babuk-derived ransomware deployed, per QUIRSO. NCSC-NL reported active exploitation of macOS Screen Sharing flaw CVE-2026-65400 (CVSS 9.8) to install Monero miners on systems with port 5900 exposed. Lazarus Group exploited Windows zero-day CVE-2026-68820 (CVSS 7.0) in Operation Dream Job, delivering ForestTiger and Troy backdoors to defense and aerospace targets. GeoServer patched a critical SQL injection flaw exploited within hours of disclosure, and Jamf analyzed the new Amnesia Stealer for macOS with live browser control.

The Hacker News·Aug 17, 17:57 UTC · 29d agoExploit / PoC in the wildCVE-2026-59310CVE-2026-65400CVE-2026-688201

78

macOS 27.0 RC (26A428)

Apple seeded the macOS 27.0 Release Candidate (build 26A428) to developers ahead of the final public release.

Apple published a Release Candidate build of macOS 27.0, labeled 26A428, on its developer release page. The notice only provides download and release-notes links and includes no security content, CVEs, or threat information. RC builds typically precede the general availability of the final operating system version.

Apple software releases·updated Sep 14, 17:00 UTC · 1d ago·first Sep 9, 17:00 UTC · 6d agoAdvisory 2 sources

25

[webapps] Probo 0.222.2 - IDOR

A public proof-of-concept exploit was released for an IDOR vulnerability in web application Probo 0.222.2.

Exploit-DB published exploit #52650 targeting Probo version 0.222.2. The flaw is an Insecure Direct Object Reference (IDOR) in this web application. No exploitation in the wild is reported in the disclosure text.

Exploit-DB·Aug 17, 00:00 UTC · Aug 17, 2026Exploit / PoC

18

visionOS 27.0 beta 8 (24M5361a)

Apple seeded visionOS 27.0 beta 8 (build 24M5361a) to developers, continuing the Vision Pro pre-release testing cycle.

Apple released visionOS 27.0 beta 8, build 24M5361a, to its developer program on August 31, 2026. The listing offers only downloads and release notes with no disclosed security fixes or CVEs. This is the eighth beta in the visionOS 27.0 cycle ahead of the stable release.

Apple software releases·Aug 31, 17:00 UTC · 15d agoAdvisory

10

NVIDIA Details BioNeMo Inference Runtime (BioIR): 2.90x Higher Boltz-2 Folding Throughput and 58.5K Residues per GPU-Hour on 8xH100

NVIDIA released BioNeMo Inference Runtime (BioIR), an open-source PyTorch-compatible library delivering 2.90x higher Boltz-2 protein-folding throughput on 8xH100 GPUs.

NVIDIA detailed BioIR, a Python library that accelerates Boltz-2, OpenFold2, and OpenFold3 structure-prediction inference on NVIDIA GPUs while preserving standard PyTorch workflows. On a matched benchmark of 1,000 human dimers on 8xH100 80GB GPUs, BioIR delivered 58.5K folded residues per GPU-hour versus 20.2K for a torch.compile baseline, a 2.90x throughput gain. BioIR already powered the AlphaFold Database expansion, generating about 31 million candidate complexes across 4,777 proteomes, with 1.81 million released as high-confidence predictions. Extrapolated to 1 million targets, estimated folding energy drops from 35 MWh to 11 MWh at 8-GPU TDP equivalents.

MarkTechPost·Sep 10, 21:57 UTC · 5d agoAI tools & infra1

25

macOS 27.0 beta 8 (26A5425a)

Apple seeded macOS 27.0 beta 8 (build 26A5425a) to developers with no security fixes disclosed ahead of the final release.

Apple released macOS 27.0 beta 8, build 26A5425a, through its developer program on August 31, 2026. The listing contains only download links and release notes with no vulnerability or CVE information. It is the eighth beta seed in the macOS 27.0 pre-release cycle.

Apple software releases·Aug 31, 17:00 UTC · 15d agoAdvisory

10

Perplexity Details Its GPU Embedding Stack: How Ivy, Tulip and ROSE Serve pplx-embed

Perplexity details its GPU embedding serving stack (Ivy, Tulip, ROSE), which reuses LLM prefill/decode kernels, CUDA graphs, and LazyTensors to cut launch overhead.

Perplexity engineers published a deep dive on the serving infrastructure behind pplx-embed, used across Perplexity Search and its API platform. The stack comprises Ivy (Rust HTTP gateway), Tulip (gRPC scheduling and batching), and ROSE (Runtime-Optimized Serving Engine), which reuses LLM prefill and decode kernels rather than running a separate embedding engine. Optimizations include whole-model CUDA graphs with lazy capture and a LazyTensor abstraction that overlaps CPU batch preparation with in-flight GPU work. Benchmarks are reported against vLLM v0.22.0 in BF16, with FlashAttention 4 generally fastest but FlashInfer 3 winning on Qwen-based models at very long sequence lengths.

MarkTechPost·Sep 6, 03:20 UTC · 10d agoAI tools & infra1

33

tvOS 27.0 RC (24J360)

Apple seeded tvOS 27.0 release candidate build 24J360 to developers ahead of the general release.

Apple released the tvOS 27.0 release candidate (build 24J360) through its developer releases feed. The listing only provides downloads and release notes, with no security content or CVE details. RC builds typically precede the public availability of the final OS version.

Apple software releases·Sep 9, 17:00 UTC · 6d agoAdvisory

12

Gemini Omni 1.1 Flash lets you build with more control

Google DeepMind released Gemini Omni 1.1 Flash, an updated model giving developers more control when building applications.

Google DeepMind announced Gemini Omni 1.1 Flash in a blog post titled 'Gemini Omni 1.1 Flash lets you build with more control.' The update targets developers building on Gemini, emphasizing greater control over model behavior. No article text was available, so technical details such as benchmarks, context window, or pricing are unknown.

Google DeepMind·Aug 27, 16:11 UTC · 19d agoModel release

45

Up to 3.2x Faster Inference with LFM2.5-DSpark

LiquidAI's LFM2.5-DSpark delivers up to 3.2x faster inference, announced via the Hugging Face blog.

LiquidAI announced LFM2.5-DSpark on the Hugging Face blog, claiming up to 3.2x faster inference. The release focuses on improved runtime performance for the LFM2.5 model family; further technical details were not available in the provided text.

Hugging Face Blog·Aug 20, 16:52 UTC · 26d agoAI tools & infra

35

deepseek-ai/DeepSeek-V4-Flash-Vision-Exp — new model trending #10 on Hugging Face

DeepSeek released DeepSeek-V4-Flash-Vision-Exp, its first experimental multimodal model, with large multimodal agent benchmark gains over V4-Flash-0731.

DeepSeek AI published DeepSeek-V4-Flash-Vision-Exp, an experimental multimodal model built on the DeepSeek-V4-Flash architecture with added visual modules and continued training. It scores 83.9 on Terminal Bench 2.1 and 36.5 on ApexBench Pass@1 versus 26.2 for DeepSeek-V4-Flash-0731, while remaining comparable to Opus-4.8 on several benchmarks. The MIT-licensed repository ships a tokenizer, OpenAI-style and TXT prompt encoding, and a minimal PyTorch inference implementation, with vLLM and SGLang deployment recipes.

Hugging Face trending models·Aug 31, 06:16 UTC · 16d agoModel release1

68

Agnes-AI/Agnes-3.0-Flash — new model trending #30 on Hugging Face

Agnes AI releases open-weight Agnes-3.0-Flash Preview, a 33B multimodal model with 262k-token context under Apache 2.0.

Agnes AI released Agnes-3.0-Flash Preview, an open-weights multimodal checkpoint with 33B parameters and a 262,144-token context window under Apache 2.0. The model supports text, image, and video understanding, tool calling, and adjustable reasoning effort. The repo clarifies this preview checkpoint is distinct from the production/API Agnes 3.0 Flash model, which uses a different configuration with a 1M-token context window. Reported reference results include IFBench 74.20 and SciCode 38.08 against peers such as Qwen3.6-35B-A3B, Kimi K2.5, and MiniMax M3.

Hugging Face trending models·Sep 11, 15:50 UTC · 4d agoModel release

45

DeepSeek v4.1 Flash

DeepSeek announced v4.1 Flash, a new model version, in a terse social media post without technical details.

DeepSeek publicly announced v4.1 Flash via its official X account. The announcement contained no benchmarks, parameter counts, or capability details, and further information was not yet available at publication time.

Hacker News · AI·updated Sep 11, 06:46 UTC · 5d ago·first Sep 10, 06:11 UTC · 6d agoModel release 2 sourcesHN 75↑ · 13 comments

50

[webapps] Grav CMS 2.0.7 - RCE

A proof-of-concept exploit for remote code execution in Grav CMS 2.0.7 has been published on Exploit-DB.

Exploit-DB lists a public proof-of-concept exploit for a remote code execution vulnerability in Grav CMS 2.0.7. The listing is for web applications and provides code defenders can use to reproduce the issue. No exploitation in the wild or CVE assignment is stated in the listing.

Exploit-DB·Sep 1, 00:00 UTC · 15d agoExploit / PoC

30

visionOS 26.6.1 (23O780)

Apple shipped visionOS 26.6.1 (build 23O780), a security and maintenance update for Vision Pro headsets.

Apple listed the visionOS 26.6.1 release (build 23O780) on August 17, 2026. The announcement contains only download and release note links, with no CVE identifiers or exploitation details disclosed in the feed text. Updates to visionOS generally bundle security patches alongside functional improvements for Apple Vision Pro.

Apple software releases·Aug 17, 17:00 UTC · 29d agoAdvisory

20

Introducing @huggingface/kernels: 200+ WebGPU Kernels for Local AI

Hugging Face released @huggingface/kernels, a library offering 200+ WebGPU compute kernels to accelerate AI inference locally in browsers.

Hugging Face introduced the @huggingface/kernels package, bundling more than 200 optimized WebGPU compute kernels for running AI workloads locally. The release targets browser-based and on-device inference, reducing reliance on server-side compute. No article body was available beyond the title, so benchmark results and supported models are not specified.

Hugging Face Blog·Sep 1, 00:00 UTC · 15d agoAI tools & infra

32

Introducing ChatGPT Images 2.5

OpenAI released ChatGPT Images 2.5, improving generation of personalized, polished images from ideas, sketches, and reference photos.

OpenAI announced ChatGPT Images 2.5, a new version of its image generation capability in ChatGPT. The update is designed to turn ideas, sketches, and reference photos into more personalized and polished images that better reflect user intent. No benchmark numbers, model sizes, or technical architecture details were disclosed in the announcement.

OpenAI News·Sep 8, 11:30 UTC · 8d agoModel release

50

Implementation of Machine Learning Workflows with NVIDIA cuML, RAPIDS, GPU Benchmarking, Explainability, Clustering, and Model Inference

Hands-on tutorial implements NVIDIA cuML and RAPIDS to GPU-accelerate scikit-learn-style ML workflows with benchmarking, clustering, and inference.

The tutorial demonstrates NVIDIA cuML as a GPU-accelerated machine learning framework, using cuml.accel to speed up unmodified scikit-learn scripts with zero code changes and the native cuML API for CuPy/cuDF interoperability. It benchmarks CPU versus GPU implementations of PCA, K-Means, nearest-neighbor search, logistic regression, random forests, and DBSCAN on datasets up to 200,000 samples with 64 features. It also builds GPU pipelines with UMAP, t-SNE, and HDBSCAN, validates GPU-generated SHAP explanations, uses the FIL library for forest inference, and covers model serialization and GPU/CPU portability.

MarkTechPost·Sep 13, 01:42 UTC · 3d agoAI tools & infra

14

[webapps] PodcastGenerator 3.2.9 - Stored XSS

A stored cross-site scripting flaw in PodcastGenerator 3.2.9 is documented with a public proof-of-concept exploit on Exploit-DB.

Exploit-DB published exploit ID 52677 targeting PodcastGenerator 3.2.9, a web application affected by stored cross-site scripting. The listing provides a proof-of-concept for the flaw but includes no CVE identifier or reports of active exploitation.

Exploit-DB·Sep 2, 00:00 UTC · 14d agoExploit / PoC1

12

ToolHive: The open-source way to run any MCP server securely

Stacklok's open-source ToolHive runs Model Context Protocol servers in isolated containers with per-request identity enforcement, audit logging and a signed registry.

ToolHive, shipped under Apache 2.0, containersizes MCP servers locally via Docker or Podman or in clusters through a Kubernetes operator, applying permissions, network filtering, and secrets management. The platform includes a Registry Server implementing the official MCP Registry API with signing and provenance verification, a Virtual MCP Server gateway with OIDC/OAuth single sign-on and OpenTelemetry traces, and a desktop Portal for one-click installs. The browser-based cloud UI is retired, so the desktop app and CLI are the supported rollout paths.

Help Net Security·Sep 7, 00:00 UTC · 9d agoAI tools & infra1

45

FlashVector: Agent for Hierarchical Model Serving Stack Optimization

FlashVector agent optimizes all layers of Unity's ad-serving stack, delivering up to 2x model-server throughput and 1.98x latency speedup in production.

FlashVector is an agentic system that optimizes performance across GPU kernels, ML framework computation graphs, model servers, and on-demand feature processing. Deployed in Unity's Vector advertising platform, it achieved up to 2x model-server throughput increase, 1.98x latency speedup, and 1.6x feature-store throughput gain. Optimizations spanned NVIDIA Triton's C++ codebase and the Python feature transformation service, demonstrating extensibility beyond single-kernel tuning.

arXiv cs.AI / cs.LG / cs.CL·Sep 15, 16:27 UTC · 21h agoAI research

30

Show HN: Pelican-bicycle alternatives (updated for 2026)

Hobbyist benchmark re-runs the pelican-bicycle SVG test on six 2026 frontier models, comparing generation time and API cost per image.

A Show HN post re-runs the classic pelican-bicycle and similar SVG generation tests across six 2026 models: GPT-6 Astra, Claude Fable 5.1, Gemini 3.8 Flash, DeepSeek V4 Pro, Qwen3.8 Max, and Fugu Ultra v2, recording wall-clock time and cost. It also lists 2025 baseline runs with ten models including Claude Sonnet 4.5, GPT-5.2 Pro, and Qwen3-VL-235B-A22B-Thinking. DeepSeek V4 Pro is consistently cheapest ($0.04-$0.10) while Qwen3.8 Max is slowest, taking up to roughly 17 minutes per generation.

Hacker News · security·Sep 14, 13:20 UTC · 2d agoAI researchHN 24↑ · 4 comments

12

[AINews] Fal’s H3 Max Live breaks the infinite videogen barrier

Fal post-trained MiniMax H3 into a 'Max' variant with 35x-faster inference, enabling faster-than-realtime AI video generation and infinite streams.

Fal post-trained MiniMax's H3 model into a 'Max' variant and optimized it for its in-house inference engine, achieving roughly 35x the speed of the official endpoint. The optimization enables faster-than-realtime video generation, demonstrated by an infinite interactive AI-generated stream productized by levels.io. The roundup also notes Meta Muse Code's general availability with an SDK, open DeepSeek-V4-Flash-Vision-Exp weights, GLM-5.3-Flash's strong agentic cost/performance rankings, and Tencent's 770B-parameter Hy4 Preview MoE with 49B active parameters.

Latent Space·Sep 1, 04:36 UTC · 15d agoAI industry

50