Stay up to date with the latest features and improvements
Credits Cover Every Feature, Long Chats Stop Crashing
The credit ledger now spans every AI feature instead of just chat, images, and video, long streamed answers stopped crashing mid-sentence, and we fixed PDF uploads, the Chrome extension, and video pricing.
ImprovedThe shared credit ledger now covers every AI feature — agentic chat, the developer API, TTS, STT, web search, and scraping — not just chat, images, and video, so usage is fungible across all of them instead of siloed
FixedLong streamed chat answers no longer crash mid-sentence with a 'Minified React error #185' card — the crash was in the browser's auto-scroll animation, and the answer was always saved correctly on the server
FixedPDF and DOCX uploads work again — a backend upgrade had been mislabeling converted documents, causing some providers to reject them outright
ImprovedChrome extension reconnected to the current backend with a dedicated per-installation sign-in and revocable developer key, instead of the older shared credential path
NewModel catalog and comparison pages rebuilt on live, sourced data — real pricing, context windows, benchmarks, and Arena rankings for the full 400+ model catalog
FixedSeedance 2.0 video generation was billing roughly 8x too little per second — pricing is now correct, and AppSumo video allowances update to new monthly tiers with existing entitlements grandfathered
FixedTrial subscription upgrades no longer fail with a Stripe configuration error when confirming a plan change
FixedFree-plan users who also belong to a paid workspace can now close the upgrade paywall instead of getting stuck with no way back into the app
ImprovedModel catalog, docs, and blog pages load faster again after restoring static caching that had regressed to slower dynamic rendering
Agent File Search, Image History, and Broken Link Fixes
A quieter week focused on fixes — Saved Agents with file search now execute properly instead of printing raw tool text, image history stops truncating at 100 items, and broken chat links fail cleanly instead of erroring out.
FixedSaved Agents with file search enabled now actually run the search and answer your question, instead of sometimes showing a raw 'searchFile' instruction and still counting the message against your usage
FixedImage history no longer stops loading at 100 items — scrolling back through a large generation library now retrieves your full history in order
FixedChat links with an invalid or missing ID now show a proper not-found page instead of throwing a server error
ImprovedPage titles and search result descriptions across the site now generate within proper length limits, and a batch of broken redirects and insecure links were fixed
Writingmate MCP Joins the Official Registry
Writingmate's MCP server is now discoverable through the official MCP Registry, we published a real guide for generating video through Claude, and we closed out an image generation bug and a usage-counting bug found in production.
NewWritingmate MCP server is now listed in the official MCP Registry, so Claude, ChatGPT, and other MCP clients can discover and connect to it without you pasting in a server address
NewNew guide on generating video through Claude — covers exact MCP setup and which video models (Seedance 2.0, Sora 2, Veo 3.1, Kling, PixVerse) it can call
FixedImage generation requests that come back malformed from the provider now fail with a clear error instead of a confusing crash or a silently dropped request
FixedUsage and quota counts no longer drift out of sync when a provider returns an invalid token count on API requests
ImprovedHomepage product demo now matches the rest of the site's visual design
A Week of Reliability Fixes Across Login, Billing, and Chat
No major new feature this week — we spent it closing out a login bug that could lock people out, a billing error on trial upgrades, and a handful of other production issues found in real user traffic.
FixedLogin no longer fails when a session cookie gets corrupted — accounts that were getting silently locked out now recover automatically instead of throwing an error
FixedUpgrading a trial subscription with no actual plan change no longer throws a billing error
FixedChat no longer errors out when a message references a broken or invalid file search ID
FixedAI agents that call tools in a loop no longer lose their final answer — you always get a response back instead of a dropped reply
FixedImage editing requests that come back incomplete from the provider now fail with a clear error instead of hanging silently
FixedVoice transcription cleanup no longer invents extra sentences or repeats phrases at the end of a transcript
ImprovedOld links to retired pages now redirect to the right section instead of showing a 404
Claude and ChatGPT can now connect to Writingmate's MCP server with a one-click sign-in instead of a manual API key, the /developers page got a full redesign, and we closed out a batch of billing and generation reliability bugs.
NewOAuth login for MCP — connect Claude, ChatGPT, or any MCP-compatible client to Writingmate with a one-click sign-in instead of pasting an API key
NewWalkthrough videos — click-to-play demo videos added to the /developers page and several free AI tools so you can see a feature in action before trying it
Improved/developers page redesigned to match the rest of Writingmate, with clearer setup steps and an up-to-date list of models included in your plan
ImprovedVideo generation failures now explain the real reason, like a moderation block or an unsupported input, instead of a generic error
ImprovedImage generation for image-only models now routes through the correct API, fixing failed generations for those models
FixedTrial subscription upgrades and cancellations now reconcile correctly with Stripe, so paid plans no longer get overwritten or stuck mid-trial
FixedDeleting an AI agent no longer times out
FixedStreaming chat requests now surface the real error, like an unavailable model, instead of a generic failure
FixedAudio transcription and image generation errors now show the actual provider reason instead of a generic server error
Chat Effort Selector and a Developer API
Pick how hard reasoning models think before they answer, connect your own agent tools through a new developer API and MCP endpoint, and get a handful of billing and reliability fixes.
NewChat effort selector — choose low, medium, or high reasoning effort for Claude and other reasoning-capable models right next to the model picker
NewDeveloper API & MCP endpoint — a new /developers page and an authenticated MCP JSON-RPC surface so agent tools like Claude Code, Cursor, and OpenCode can call chat, image, and video generation directly
ImprovedOnboarding checklist steps can now be skipped individually instead of only being dismissed all at once, including a new prompt to invite a teammate
ImprovedReasoning effort controls now reflect each model's real capability data, so the selector shows up correctly instead of being guessed from the provider name
FixedTrial subscription cancellation now works correctly instead of silently failing
FixedAppSumo lifetime deal credits reset properly on your monthly anniversary date
FixedLong conversations with Claude models no longer fail with a cache-overflow error
FixedMath expressions in markdown now render correctly instead of showing raw syntax
FixedVoice dictation no longer swaps short spoken commands for unrelated words from your custom vocabulary list
Chat with Any File, Seedance 2.0 Video & Usage Dashboard
No more 'file too large' errors. Upload any document and ask questions about it. Plus new video generation and a way to track your usage.
NewChat with files of any size — no more 'file too large' errors. Ask questions about long documents and the AI reads the relevant sections for you
NewSeedance 2.0 video — create cinematic videos with built-in audio, realistic physics, and camera control
NewUsage dashboard — see how many messages you've used today, which models you use most, and your usage over time
NewOpenAI-Compatible API — connect coding agents like OpenCode, OpenClaw, and Aider to your Writingmate account
ImprovedMore generous daily limits for AppSumo users — you can now use up to 30 Pro and 5 Ultimate messages per day on Tier 1
ImprovedAttach files to any model — PDFs, text files, and documents work everywhere, not just on select models
FixedFollow-up messages now stay in the same conversation instead of creating a new one
AI Image Editing & OpenAI-Compatible API
Edit your AI-generated images directly in the viewer, and connect any OpenAI-compatible tool to Writingmate.
NewEdit AI images after generation — paint over the area you want to change, describe the edit, and the AI updates only that region
NewOpenAI-compatible API — use Writingmate from OpenCode, Aider, Cursor, or any tool that supports custom OpenAI endpoints
NewDeveloper keys — create API keys in Settings to use Writingmate programmatically
NewDownload button now visible on hover in the image gallery (no more digging through menus)
ImprovedImage library loads significantly faster with parallel queries and pagination
New Models Appear Automatically
When a new frontier model launches, it now shows up in your model list automatically.
NewNew frontier models from OpenAI, Anthropic, Google, and others appear automatically when released
ImprovedModels load instantly — no more waiting for the model list to populate
ImprovedReduced token consumption with cached prompt discounts, saving you messages on repeated conversations
AppSumo Lifetime Deal & Security
Lifetime deal available on AppSumo with credit pools, plus improved login security.
NewAppSumo lifetime deal with tiered credit pools for Pro, Ultimate, Image, and Video
NewBYOK (Bring Your Own Key) — add your OpenRouter API key for unlimited access to all models at cost
NewAnti-bot protection with Cloudflare Turnstile on email login
FixedBlocked disposable email domains used by bot farms
Projects & Media Sharing
Organize your work into projects and share your AI creations publicly.
NewProjects — organize chats into projects with file sources, instructions, and sidebar navigation (like ChatGPT Projects)
NewShare AI-generated images and videos publicly with a link
ImprovedImage and video generation UX improvements across the platform
Onboarding Wizard & Creator Program
New personalized setup flow and a way for content creators to earn with Writingmate.
New3-step onboarding wizard with personalized model recommendations based on your use case
NewCreator program — earn $2-5 per 1,000 views creating Writingmate content on YouTube
NewCancel your subscription directly from Settings without visiting Stripe
NewDedicated image generation page at /text-to-image with model selector
ImprovedFile uploads now show friendly names (PDF, DOCX) instead of MIME types
40% Off Annual Ultimate & Dynamic Model Catalog
Permanent deal on the Ultimate annual plan and a smarter model catalog.
New40% off annual Ultimate plan — $432/year instead of $720
NewModel catalog updates automatically when new models are added
Android App (PWA)
Install Writingmate as an app on your Android device.
NewInstall Writingmate as a native-feeling app on Android via your browser
ImprovedFaster page loads with service worker caching
Blog, Model Comparisons & Side-by-Side Chat
Compare AI models head-to-head and read in-depth articles about new releases.
NewCompare responses from multiple AI models side by side on desktop
NewModel comparison pages — see benchmarks, pricing, and capabilities at a glance
NewBlog with articles about GPT-5.2, Gemini 3 Flash, Nano Banana Pro, and more
NewSmart rate limit handling — when you hit a limit, choose a fallback model instantly
ImprovedFile upload now disabled for models that don't support files (prevents confusing errors)
Sora 2 & VEO 3 Video Generation
Generate AI videos with the latest models and manage your subscription more easily.
NewSora 2 and VEO 3 video generation pages with how-to guides
NewNano Banana Pro image generation with 20 showcase examples
NewCancel subscription wizard — see your billing details and cancel without leaving Writingmate
NewAI search buttons on all FAQ sections — ask ChatGPT, Perplexity, or Google about any question
ImprovedPricing page redesigned with model badges, testimonials, and payment methods
Mobile Experience & Table Improvements
Enhanced mobile usability and added powerful table features for better data management.
NewTables now display in spreadsheet style with grid borders and alternating row colors
NewExport any table to CSV with one click (hover to reveal download button)
ImprovedArtifact panel now overlays as a sheet on mobile instead of squeezing content
FixedFixed mobile view overflows and scrolling issues
FixedFixed chat auto-scroll behavior when new messages arrive
