LinkWord
Home
Directory
Articles
AI models
Tools
Pixel Plaza
Settings
ContactRSSFriend linksSubmit site
Privacy Policy陕ICP备2025083618号-2

Hot channels

Developer ToolsAI ToolsProductivity ToolsSecurity ToolsDesign Resources

LinkWord · Find great sites and reads

Discover AI sites and tools worth bookmarking

Search when you know what you want—or scroll on: articles, tools, and picks are just below.

You don’t need the whole site at once. Start wherever feels natural.

Browse the directoryRead a storyTry a tool

LinkWord

What is LinkWord?

LinkWord (linkwordx.site) is an AI directory and navigation-style site: it combines an AI tools directory, website recommendations, tech news, curated articles, and a large set of small online utilities—so you can discover great bookmarks and jump to what you need with search.

Search the directory from the home page, browse categories, read the latest articles, or open the toolbox for privacy-friendly, in-browser helpers for links, text, and data—no install or login required.

Browse the AI & website directoryRead tech news & curated articlesOpen the online toolbox

Editor’s picks · what’s trending

Not sure what to search yet? Start with what we’re featuring, then peek at what others have been opening lately—you might find your next favorite bookmark.

Open link

People are clicking these lately

Sorted by clicks

1Site icon for “GitHub”GitHub5 clicks2Site icon for “dverso Laundry — Background Removal”dverso Laundry — Background Removal3 clicks3CChuangge Toolbox — A collection of free small tools, ready to use by simply opening the webpage, with files processed as much as3 clicks4Site icon for “Railway”Railway2 clicks5Site icon for “Gamma — AI Presentations & Website Builder”Gamma — AI Presentations & Website Builder2 clicks6Site icon for “Claude”Claude2 clicks7Site icon for “Snyk”Snyk2 clicks8Site icon for “Yingjiesheng”Yingjiesheng2 clicks9Site icon for “GitLab”GitLab1 clicks10Site icon for “Travis CI”Travis CI1 clicks
Browse directory

Worth a read

Cover image for “DeepSeek V4 Pro Gets a Permanent 75% Price Cut”

DeepSeek V4 Pro Gets a Permanent 75% Price Cut

DeepSeek permanently cuts V4 Pro API pricing by 75%, output drops to $0.87/M tokens (≈¥5.92). Cache-hit pricing across all models slashed to one-tenth.

Cover image for “2026-05-22 Picks: Freenet, Runtime, Agent.email”

2026-05-22 Picks: Freenet, Runtime, Agent.email

Today's HN picks: Freenet's decentralized platform relaunch, Runtime's team control plane for AI coding agents, and Agent.email's email service built for AI agents.

Cover image for “2026-05-21 Open-Source Tool "Remove-AI-Watermarks" Hits HN Front Page: The AI Watermark Arms Race Is On”

2026-05-21 Open-Source Tool "Remove-AI-Watermarks" Hits HN Front Page: The AI Watermark Arms Race Is On

An open-source tool called remove-ai-watermarks hit the HN front page, capable of removing Gemini's visible watermark, SynthID's invisible watermark, and C2PA metadata in one command. The same day, OpenAI adopted SynthID — the AI watermark arms race is officially on.

AI model leaderboard

Top 5 preview per category—chat, image, and video.

Bar chart comparison

Score
RankModelProvider
ScoreEditorial composite score; higher ranks higher
Context (K)Context window size (thousand tokens)
Input $Input token price (USD per 1M tokens)
Output $Output token price (USD per 1M tokens)
MMLUMassive Multitask Language Understanding accuracy (%)
HumanEvalCode generation benchmark pass rate (%)
EloArena Elo from human preference battles; higher is stronger
1
G
GPT-5.5
—98.0105053091.5091.401420
2
G
Gemini 3.5 Pro
Google DeepMind96.010001.5099189.501400
3
C
Claude Opus 4.7
—93.010005259192.501400
4
G
Gemini 3.5 Flash
—90.010491.50992.3086.801370
5
G
GPT-5.5 Instant
OpenAI88.09220.75389.5088.201350
View full leaderboard

Handy tools

A few crowd favorites live here; the full toolbox has 50+ more, all running locally in your browser.

UTM link builder and parser

Add or inspect utm_source, utm_medium, utm_campaign, and more for campaign tracking.

QR code generator

Create QR codes locally: size and colors, PNG/SVG export, WiFi/vCard templates, optional center logo.

JSON / YAML tools

Format or minify JSON, or convert between JSON and YAML.

AI token calculator

OpenAI models via tiktoken; China models via common heuristics and cl100k/o200k proxies for rough cost and context planning.

Browse all tools

Three ways to use this site—mix and match

Find sites, catch up with reads, or knock out a quick task in the browser. It all works best with the search box above.

Land on the right site

Filter with keywords—or let AI understand what you need—so you spend less time bouncing between tabs.

Reads that go a little deeper

Tech, AI, and product pieces for a commute scroll or a longer weekend read.

Little tools, no install

Links, text, data, color helpers, and more—open, use, close. No account required.

Longer reads

Fresh articles and takes—subscribe with RSS if you want updates in your reader.

View all →
Cover image for “DeepSeek V4 Pro Gets a Permanent 75% Price Cut”New

DeepSeek V4 Pro Gets a Permanent 75% Price Cut

DeepSeek permanently cuts V4 Pro API pricing by 75%, output drops to $0.87/M tokens (≈¥5.92). Cache-hit pricing across all models slashed to one-tenth.

5/22/2026

Cover image for “2026-05-22 Picks: Freenet, Runtime, Agent.email”New

2026-05-22 Picks: Freenet, Runtime, Agent.email

Today's HN picks: Freenet's decentralized platform relaunch, Runtime's team control plane for AI coding agents, and Agent.email's email service built for AI agents.

5/22/2026

Cover image for “2026-05-21 Open-Source Tool "Remove-AI-Watermarks" Hits HN Front Page: The AI Watermark Arms Race Is On”New

2026-05-21 Open-Source Tool "Remove-AI-Watermarks" Hits HN Front Page: The AI Watermark Arms Race Is On

An open-source tool called remove-ai-watermarks hit the HN front page, capable of removing Gemini's visible watermark, SynthID's invisible watermark, and C2PA metadata in one command. The same day, OpenAI adopted SynthID — the AI watermark arms race is officially on.

5/21/2026

Cover image for “2026-05-21 Picks: yapsnap, Map of Metal, DOS Zone”

2026-05-21 Picks: yapsnap, Map of Metal, DOS Zone

Today's curation from Hacker News: yapsnap CPU video transcriber, Map of Metal interactive music genre explorer, and DOS Zone retro gaming portal. Three unique finds worth bookmarking.

5/21/2026

Cover image for “Artificial Analysis: Check Here Before Picking Your AI Model”

Artificial Analysis: Check Here Before Picking Your AI Model

Artificial Analysis is one of the most comprehensive AI model comparison platforms. It evaluates every major model under the same methodology—intelligence, coding ability, speed, and pricing. This article covers its key features and recommends pairing it with our own AI Model Rankings for a more complete picture.

5/20/2026

Cover image for “Cursor Composer 2.5: Built on Kimi K2.5, Performance Rivals Frontier Models”

Cursor Composer 2.5: Built on Kimi K2.5, Performance Rivals Frontier Models

Cursor releases Composer 2.5, built on Kimi K2.5 open-source checkpoint, rivaling frontier models at 1/10 the cost. Partnering with SpaceXAI on Colossus 2.

5/20/2026

Cover image for “Google Gemini Omni: Where Reasoning Meets Creation, Video Editing Goes Conversational”

Google Gemini Omni: Where Reasoning Meets Creation, Video Editing Goes Conversational

Google DeepMind launches Gemini Omni — a unified multimodal model for conversational video editing with multi-turn consistency and physics understanding.

5/20/2026

Cover image for “Google Releases Gemini 3.5 Flash: Frontier Intelligence Meets Action”

Google Releases Gemini 3.5 Flash: Frontier Intelligence Meets Action

Google DeepMind releases Gemini 3.5 Flash at I/O, combining frontier intelligence with action — supporting agentic workflows and tool calling at 4x speed.

5/20/2026

Cover image for “Claude Is Quietly Overtaking ChatGPT: Revenue Growth, Stainless Acquisition, and the Vatican”

Claude Is Quietly Overtaking ChatGPT: Revenue Growth, Stainless Acquisition, and the Vatican

Claude has surpassed ChatGPT in enterprise revenue and new customer acquisition. Anthropic acquires API tooling company Stainless, co-founder to co-present AI encyclical with Pope Leo XIV. The enterprise AI strategy is working.

5/19/2026

Cover image for “Qwen Ecosystem Explodes: From 3.5 to 3.7, Alibaba's Open-Source LLM Is Now Top Tier”

Qwen Ecosystem Explodes: From 3.5 to 3.7, Alibaba's Open-Source LLM Is Now Top Tier

From Qwen 3.5 to 3.7's rapid iteration cycle, Qwen3.6-27B's flagship-level coding (993 pts HN), 27B parameters competing with much larger models, and community tool Orthrus delivering 7.8× throughput boost.

5/19/2026

Cover image for “MTP (Multi-Token Prediction) Hits llama.cpp: Local Inference Speed Doubles”

MTP (Multi-Token Prediction) Hits llama.cpp: Local Inference Speed Doubles

MTP (Multi-Token Prediction) lands in llama.cpp, doubling local inference speed. Qwen 3.x benefits most due to native MTP support. Community tests show Qwen3.5-27B hitting 207 tok/s on RTX 3090.

5/19/2026

Cover image for “Semble: Code Search That Uses 98% Fewer Tokens”

Semble: Code Search That Uses 98% Fewer Tokens

Semble is an open-source code search tool for AI agents that uses semantic search instead of grep+read, achieving 98% fewer tokens. Indexes repos in ~250ms, searches in ~1.5ms, runs entirely on CPU. MCP-compatible, works with Claude Code and Cursor. NDCG@10 of 0.854, 200x faster indexing than Transformer-based approaches.

5/19/2026

Cover image for “Needle: Distilling Gemini's Tool Calling into a 26M Parameter Model”

Needle: Distilling Gemini's Tool Calling into a 26M Parameter Model

Cactus Compute distills Gemini 3.1's tool calling into Needle, a 26M parameter model that beats models 10x its size on function calling. Open source with 2100+ GitHub stars.

5/18/2026

Cover image for “Is AI Really Taking Our Jobs? What the Bloomberg Data Says”

Is AI Really Taking Our Jobs? What the Bloomberg Data Says

Bloomberg reports US AI-exposed jobs are shrinking. BLS data shows 0.2% drop in 18 occupations vs 0.8% overall growth. But is AI really to blame?

5/18/2026

Cover image for “I Turned Minecraft Server Operations into a Claude Code Skill - Now Open Source”

I Turned Minecraft Server Operations into a Claude Code Skill - Now Open Source

A Claude Code Skill covering both Java and Bedrock Edition Minecraft servers, handling everything from setup to troubleshooting. Includes an RCON client built with pure Python standard library, requiring zero external dependencies.

5/16/2026

Cover image for “Tencent Hunyuan Hy3 Preview Tops OpenRouter Charts: The First Model After Yao Shunyu Reconstruction”

Tencent Hunyuan Hy3 Preview Tops OpenRouter Charts: The First Model After Yao Shunyu Reconstruction

Tencent Hunyuan Hy3 Preview tops OpenRouter charts within two weeks of release. The 295B-parameter MoE model, the first since Yao Shunyu infrastructure reconstruction, delivers 40% improved inference efficiency and enhanced Agent capabilities.

5/16/2026

Cover image for “Anthropic Adds "Dreaming" Feature to Claude Managed Agents”

Anthropic Adds "Dreaming" Feature to Claude Managed Agents

Anthropic announced "dreaming" for Claude Managed Agents, a feature that periodically reviews sessions across agents to identify and store important patterns.

5/16/2026

Cover image for “GPT-5.5 Instant vs. Grok 4.3: Who Is the Current King of Value?”

GPT-5.5 Instant vs. Grok 4.3: Who Is the Current King of Value?

In May 2026, OpenAI released GPT-5.5 Instant as the default model for ChatGPT, while xAI launched Grok 4.3 to disrupt the market with aggressively low pricing. This article provides an in-depth comparison across four dimensions—performance benchmarks, pricing strategy, Agent capabilities, and security—to help you determine which model best fits your use case.

5/7/2026

Cover image for “GPT-5.5 Instant Officially Released: Double the Speed, Uncompromised Reasoning, Finally Here for Developers”

GPT-5.5 Instant Officially Released: Double the Speed, Uncompromised Reasoning, Finally Here for Developers

OpenAI officially launches the GPT-5.5 Instant model, achieving sub-200ms latency and 60% cost reduction while retaining GPT-5.5's core reasoning power. This article helps you determine if switching is worthwhile across three dimensions: benchmark data, use cases, and a migration guide.

5/7/2026

Cover image for “DeepSeek V4 vs. GPT-5.5: Comprehensive Benchmark Data Comparison – Open-Source Flagship vs. Closed-Source Champion Performance Showdown”

DeepSeek V4 vs. GPT-5.5: Comprehensive Benchmark Data Comparison – Open-Source Flagship vs. Closed-Source Champion Performance Showdown

Based on the latest officially released benchmark data, this article presents a comprehensive quantitative performance comparison between the DeepSeek V4 series (Pro Max and Flash Max) and OpenAI’s GPT-5.5. Featuring detailed data tables covering over 20 key evaluation metrics, we clearly illustrate the specific strengths and gaps between the two models across dimensions including knowledge, reasoning, programming, math, and agentic capabilities, providing developers and enterprises with a precise, authentic data foundation for technology selection.

4/24/2026

Cover image for “GPT-5.5 Officially Released: A Comprehensive Upgrade for the OpenAI Agent Era”

GPT-5.5 Officially Released: A Comprehensive Upgrade for the OpenAI Agent Era

On April 23, 2026, OpenAI officially launched its next-generation flagship model, GPT-5.5—a major update just seven weeks after GPT-5.4. Positioned as "a new intelligence tier for real-world work," GPT-5.5 delivers significant breakthroughs in agent coding, computer use, knowledge work, and early-stage scientific research. Compared to GPT-5.4, the new model handles more complex tasks with lower token consumption while maintaining comparable latency, marking a pivotal shift for AI from conversational tools to autonomous executors.

4/24/2026

Cover image for “ChatGPT Images 2.0 Officially Released: When Image Models Learn to "Think," AI Design Enters the Productivity Era”

ChatGPT Images 2.0 Officially Released: When Image Models Learn to "Think," AI Design Enters the Productivity Era

On April 22, 2025 (local time), OpenAI officially launched its next-generation image generation model, ChatGPT Images 2.0 (codenamed gpt-image-2). This is not merely an improvement in resolution or style optimization, but a qualitative leap from a "rendering tool" to a "visual system." It introduces reasoning capabilities into image generation for the first time, supporting pixel-level text rendering and precise multi-language output, while also enabling web browsing, autonomous planning, and the generation of up to eight coherent images in a single pass under Thinking mode. From ranking first in three categories on the Arena leaderboard to OpenAI's self-deprecating Chinese comic demo "Steadily Catching You," Images 2.0 is redefining the ceiling of AI image generation.

4/23/2026

Cover image for “Link Think: One-Click Model Switching, Unlocking the New Era of Cost-Effective AI Integration”

Link Think: One-Click Model Switching, Unlocking the New Era of Cost-Effective AI Integration

This article provides an in-depth review of Link Think, an AI capability aggregation platform. It details how this 'AI Gateway' leverages unified compatible APIs, multi-model aggregation, intelligent routing, and granular cost management to help developers and enterprises seamlessly, reliably, and cost-effectively integrate leading LLMs like OpenAI and Claude—achieving reduced costs, enhanced efficiency, and robust high availability.

4/22/2026

Cover image for “Kimi K2.6 Officially Released: Major Breakthroughs in Long-Context Coding and Multi-Agent Cluster Capabilities”

Kimi K2.6 Officially Released: Major Breakthroughs in Long-Context Coding and Multi-Agent Cluster Capabilities

Moonshot AI officially released and open-sourced its latest flagship model, Kimi K2.6, on April 20. The model achieved significant upgrades in code generation, long-horizon task execution, and multi-agent cluster capabilities, performing on par with or surpassing mainstream closed-source models like GPT-5.4 and Claude Opus 4.6 across multiple benchmarks. This marks a new milestone for domestic large models in complex software engineering and automated task processing.

4/21/2026

Load more

Where to next?

If you already know what you need, head to directory search. If you’d rather read or play with a tool, the doors below are open.

Search the directoryBrowse articlesOpen the toolbox
Suggest a site / say hiSubscribe via RSS

Follow on WeChat

WeChat account QR code