Articles by Writingmate

Stay updated on the latest news about AI. Explore insightful articles and tips.

Inkling model listing in the Writingmate AI model directory showing its 975B MoE architecture and 1M-token context window

models

July 18, 2026

Artem Vysotsky

Inkling Is on Writingmate: Testing Thinking Machines Lab's First Open-Weight Model for Agentic Coding and Tool Use

Inkling is Thinking Machines Lab's first open-weight model — a 975B-parameter MoE with 41B active parameters built for agentic coding and tool use. I ran it through the same test suite I used on Kimi K3 and Grok 4.5 to see where it actually holds up.

All Articles

Explore our collection of articles covering AI tools, productivity tips, and industry insights.

Kimi K3 release card for Writingmate users

Kimi K3: Moonshot's 2.8T Open Flagship — Now on Writingmate and in OpenCode via Our API

Moonshot AI's Kimi K3 is the largest open-weight model ever released, and it is live on Writingmate. Use it in chat, or plug it into OpenCode through our OpenAI-compatible API.

Gemini 3 Flash Preview listed in the Writingmate model catalog next to other agentic coding models

Gemini 3 Flash Preview Is on Writingmate: Testing Google's Fast Thinking Model for Agentic Coding and Everyday Chat

Gemini 3 Flash Preview promises near-Pro reasoning at Flash-level speed and cost. I put that claim through an agentic coding, tool-use, and multi-turn chat test plan to see exactly where it holds up and where it quietly falls apart.

product-updates

What Shipped This Week: OAuth Logins for MCP, a Rebuilt Developer Page, and a Round of Reliability Fixes

Claude and ChatGPT can now sign in to Writingmate's MCP server with one click instead of an API key, the /developers page got rebuilt, and we closed out a string of billing and generation bugs.

Split-screen illustration of two AI writer personas handing off a fantasy story mid-scene, representing Aion-3.0's multi-model storytelling system

Aion-3.0 Is on Writingmate: Testing AionLabs' New Multi-Model System for Roleplay and Long-Form Storytelling

Aion-3.0 doesn't write like a single model — it's a relay team of specialized models handing a story off mid-scene. I ran the same long-form roleplay prompts through it, GLM 5.2, DeepSeek V4 Pro, and its own predecessor Aion-2.0 to see if the hand-off actually pays off.

Kimi K3 release card for Writingmate users

Kimi K3 Is on Writingmate: Where It Fits for Long-Context Professional Work

Kimi K3 is available in Writingmate. Here is how to evaluate it for long-context research, structured drafting, and professional reasoning work.

Step-by-step screen sequence showing a custom AI agent being created with a name, instructions, knowledge files, and connected tools

How to Create an AI Agent in 2026: A Step-by-Step Walkthrough (No Code Required)

Every "how to build an AI agent" guide stays theoretical. This one is a literal click-by-click walkthrough — name it, write instructions, add files, connect tools, and test it — so you finish with a working agent, not just a framework.

MCP hub diagram connecting external apps, users, and data sources through one server

How to Connect ChatGPT, Codex, or Claude to 600+ AI Models with Writingmate MCP

Connect Codex or Claude to Writingmate MCP, call hundreds of AI models from one assistant, and compare their answers with a repeatable evaluation prompt.

Split-screen graphic comparing a watermarked, low-resolution free AI video clip against a clean, high-resolution paid AI video clip

Free Text-to-Video AI Tools in 2026: What You Actually Get Without Paying (and When It Stops Being Worth It)

Every free AI video generator promises no cost. Here's what "free" actually restricts in 2026 — watermarks, resolution caps, disappearing credits — and the real signal for when paying finally makes sense.

Grok 4.5 release card showing the model listed in the Writingmate catalog

Grok 4.5 Is on Writingmate: Testing SpaceXAI's New Model for Agentic Coding and Tool Use

Grok 4.5 just launched with a 1.5T-parameter MoE architecture and an 83.3% Terminal-Bench 2.1 claim. Here's what happened when I actually tested it on agentic coding and tool-use tasks.

Ready to experience the power of AI?

Access 200+ AI models, custom agents, and powerful tools - all in one subscription.

Start 3-day free trial