Tech news & AI articles - LinkWord
AI and tech news, product notes, and practical tutorials—curated long reads updated regularly.
GPT-6 Astra, K2 Horizon Open Fleet, Qwen 3.8 on Cerebras: Triple AI Model Release | 2026-09-04
OpenAI releases GPT-6 Astra (99.9% on ARC-AGI-3, $10/$50 API pricing), IFM open-sources six-model K2 Horizon fleet (Apache 2.0), Cerebras launches Qwen 3.8 27B at 1,500 tokens/s
9/4/2026
10 views
Gemini 3.8 Flash, Qwen3.8-Max-0902, World Labs Atlas: AI Model Triple Release on September 2 | 2026-09-03
Three major releases on September 2: Google Gemini 3.8 Flash series targets long-horizon software engineering, Qwen3.8-Max-0902 claims Code Arena crown, World Labs Atlas redefines multimodal world models.
9/3/2026
11 views
GLM-5.3 Open-Weight Release: Post-Training Driven Coding and Cyber Capabilities | 2026-08-29
Zhipu AI released GLM-5.3 model weights on Hugging Face on August 28. The model shares its base with GLM-5.2, with all improvements from post-training. Coding capability improved 50%, cyber vulnerability discovery saw unexpected breakthroughs, and multiple benchmarks hit open-source SOTA.
8/29/2026
21 views
Z.ai Confirms Ox Alpha: Stealth GLM Model with 1M Context Outperforms GPT-5.6 on Coding | 2026-08-27
Z.ai confirmed its stealth model Ox Alpha is part of the GLM series, featuring a 1M-token context window and solving 8/10 coding tasks, outperforming Fable 5 and GPT-5.6-sol. Weights will be released openly.
8/28/2026
13 views
GLM-5.3-Flash and Qwen3.8-Flash-Next Released on Same Day: Open Source Enters 320B MoE Era | 2026-08-27
On August 26, Z.ai and Alibaba Tongyi released major open-source models on the same day. GLM-5.3-Flash with 320B parameters and MIT license becomes the largest open-source MoE model; Qwen3.8-Flash-Next previews Qwen4 architecture at 1/9 the training cost of its predecessor.
8/27/2026
8 views
OpenAI Jalapeño Inference Chip: 700W Outperforms Nvidia Blackwell | 2026-08-26
OpenAI unveiled Jalapeño, a custom inference chip consuming 700W that delivers 1.9x throughput vs Nvidia Blackwell. Meanwhile, Alibaba previews Qwen 3.8-Flash-Next (125B MoE, 6B active), hinting at the Qwen4 architecture.
8/26/2026
20 views
Alibaba Cloud Wan3.0 Video Generation Model Released: 30s Native Duration, Document Input, 12-Language Text Rendering | 2026-08-25
Alibaba Cloud released video generation model Wan3.0 on August 24, supporting native 30-second video, document format input, up to 20 reference assets, and 12-language text rendering. API pricing starts at ¥0.3/sec for 480P, with a limited-time 30% discount.
8/25/2026
14 views
GLM-5.3: Zhipu AI's Coding Model Leaps Forward with Unexpected Cyber Capabilities | 2026-08-20
Zhipu AI released GLM-5.3, achieving a 50% coding improvement and open-source SOTA on cyber vulnerability discovery through post-training RL scaling alone.
8/20/2026
17 views
Qwen3.8-27B Open Weights Released: 27B Model Beats Claude Opus 4.6 Max on Key Coding Benchmarks | 2026-08-15
Alibaba open-sourced Qwen3.8-27B on Aug 14: a 27B dense multimodal model with native image/video understanding and 262K context (up to 1M). It beats Claude Opus 4.6 Max on several coding benchmarks (SWE-bench Pro 61.7, DeepSWE 42.2) and runs on consumer hardware. DeepSeek also shipped V4-Pro GA plus peak/off-peak pricing.
8/15/2026
21 views
Gemini 3.7 Flash, DeepSeek Harness, GPT-5.6 Sol Ultrafast, Mistral OCR 4.1: AI Roundup | 2026-08-14
Four AI releases landed on August 13: Google's Gemini 3.7 Flash at half the price, DeepSeek's open-source Harness agent framework (MIT, everything is a plugin), OpenAI and Cerebras' GPT-5.6 Sol Ultrafast tier (up to 750 tokens/s), and Mistral OCR 4.1.
8/14/2026
11 views
Grok 4.6, DeepSeek V4 Pro 0813, and Open-Weight Qwen3.8-2.4T: Three Major AI Releases | 2026-08-13
Three major releases in 24 hours: xAI's Grok 4.6 matches GPT-5.6 Sol on the AA Intelligence Index (61), DeepSeek ships the official V4 Pro 0813 with much stronger agent capabilities, and Alibaba Qwen open-sources the 2.4T-parameter Qwen3.8-2.4T-A95B — the largest open-weight model ever.
8/13/2026
29 views
NVIDIA Launches Nemotron 3.5 Lightning: Open 30B Agentic Model with 4x Faster Output | 2026-08-12
On August 11, NVIDIA released Nemotron 3.5 Lightning, an open 30B (3B active) MoE agentic model with a Mamba-2 + MoE + Attention hybrid architecture and up to 1M token context, plus the open-source NeMo Switchyard routing library. NVIDIA claims up to 4x faster output and ~30% faster agentic task completion. Available on Hugging Face, ModelScope, OpenRouter, and as a NIM microservice; runs locally on RTX PCs and DGX Spark under the commercial-friendly OpenMDW-1.1 license.
8/12/2026
15 views