Changelog

Follow new updates and improvements to allmates.ai.

May 29th, 2026

New

Anthropic has officially released Claude Opus 4.8, a direct upgrade that replaces the previous 4.7 version. While it maintains the same architecture and pricing, this release introduces key optimizations in precision, consistency, and reliability for your daily workflows.

To ensure your team is always working with the most reliable intelligence, we highly recommend migrating all of your Mates currently configured on older Opus versions (4.6 or 4.7) to this new 4.8 standard today.

🔍 What’s New?

Claude Opus 4.8 focuses on delivering stable, higher-quality reasoning for enterprise knowledge work:

  • 🧠 Sharper Judgment: The model is more precise and displays greater honesty about its progress. It is significantly more likely to flag uncertainty rather than making unsupported claims.

  • 📊 Consistent Document Analysis: It delivers improved consistency and reasoning quality when analyzing large files, reports, and complex spreadsheets.

  • 💰 Same Base Pricing: You get a more reliable and polished model with absolutely no changes to your token costs compared to Opus 4.7 ($5/M input tokens, $25/M output tokens).

  • ⚡ Clean Migration: Unlike previous upgrades, Opus 4.8 does not introduce any tokenizer changes, meaning your existing prompts will work seamlessly with no token penalty.

✨ Why You Should Upgrade

  • Seamless Transition: There is no downtime or prompt re-tuning required. Your Mates will simply provide more reliable, factual responses out of the box.

  • Higher Quality Outputs: Your business analyses, executive summaries, and data interpretations benefit from Anthropic's most refined reasoning engine to date.

  • Keep Your Workspace Current: Upgrading ensures your organization is leverage-testing the latest stable production model available on the market.

🛠️ How to Migrate Your Mates in 1 Click

Upgrading your virtual team is quick and simple:

  1. Navigate to your workspace and open Manage Mates.

  2. Select the Mate you wish to update.

  3. Under the model configuration settings, change Claude Opus 4.6 or Claude Opus 4.7 to Claude Opus 4.8.

  4. Click Save / Update.

Your Mate is now instantly equipped with the latest 4.8 engine!

Take a few moments to switch your Mates today and enjoy a more reliable, refined collaborative experience.

Happy collaborating,
— The allmates.ai Team

May 20th, 2026

New

We are thrilled to announce that Google's brand-new model, Gemini 3.5 Flash, is now live and fully integrated into the allmates.ai platform! Released just today, this state-of-the-art model is set to become the ultimate workhorse for your enterprise workflows, ending the compromise between speed, power, and cost.

🔍 What’s New?

Gemini 3.5 Flash brings "Pro-grade" reasoning capabilities to the Flash family, combining lightning-fast latency with advanced cognitive performance. You can now equip your Mates with this model to unlock:

  • Adaptive Thinking Mode: The model dynamically adjusts its reasoning depth based on task complexity, ensuring you get rapid responses for simple queries and deeply analyzed answers for complex problems.

  • Massive 1-Million Token Context: Natively process massive datasets, thousands of lines of code, or highly detailed enterprise documents in a single interaction.

  • Elite Coding & Agentic Performance: Optimized specifically for multi-step workflows, tool execution, and complex reasoning, making your Mates more autonomous than ever.

  • Immediate Multi-Provider Access: Choose how you connect! Gemini 3.5 Flash is immediately available on allmates.ai via Google Vertex AI, Google AI Studio, and OpenRouter.

❤️ Why You’ll Love It

  • 💥 4x Cheaper than Flagship Models: Gemini 3.5 Flash serves as an incredible, cost-effective alternative to heavyweights like Claude Opus 4.7 and GPT 5.5—delivering near-premium reasoning and coding capabilities for about 4 times less!

  • 🎯 The Perfect Middle Ground: While it is 3 times more expensive than the entry-level Gemini Flash 3.0 (which remains fully available on our platform for lightweight, high-volume tasks), Gemini 3.5 Flash unlocks a completely new tier of intelligence and context handling that simple models cannot reach.

  • ⚡️ Unmatched Speed & Scalability: Get responses in seconds, allowing you to scale automated customer support, real-time data extraction, and development workflows without breaking the bank.

🛠️ How to Get Started

Deploying Gemini 3.5 Flash for your Mates is simple and takes less than a minute:

  1. Navigate to your organization Settings > Models Management.

  2. Activate Gemini 3.5 Flash using your preferred provider (Google Vertex AI, Google AI Studio, or OpenRouter).

  3. Go to Manage Mates, select your Mate, and attach the new model under the Tools/Model tab.

We can't wait to see the complex problems you'll solve and the highly efficient agentic teams you'll build with this new powerhouse!

Happy collaborating!

— The allmates.ai Team

April 29th, 2026

New

We are thrilled to announce a massive leap forward for your creative workflows on allmates.ai! The highly anticipated gpt-image-2 model from OpenAI is now available in your Tool Store. Say goodbye to the days of misspelled text and compromised details—this next-generation model delivers breathtaking photorealism, flawless typography, and advanced visual reasoning.

Whether you need to create brand-new visuals from scratch or edit existing images with surgical precision, your Mates just got a major creative superpower.

🔍 What’s New?

  • 📏 4K Resolution & Flexible Sizes: Generate everything from perfect social media squares to stunning 4K landscapes (up to 8.29 million pixels) to fit any project requirement.

  • ✍️ Flawless Multilingual Text: Achieve near-perfect text rendering in your images across multiple languages (including Latin, Chinese, Japanese, Arabic, and more). Logos, slogans, and UI text are finally crisp and accurate!

  • 🧠 Advanced Visual Reasoning: The model now features a "thinking mode" that understands spatial logic and complex compositions, making it the ultimate tool for infographics, diagrams, and multi-object scenes.

  • ✂️ High-Fidelity Editing: Edit existing images or blend concepts seamlessly. Your subject's identity, brand logos, and core elements are automatically preserved by default—no complex tweaking required.

  • 🎚️ Granular Quality Control: Choose between low for fast, cost-effective prototyping, medium for balanced results, or high for premium, final-delivery renders.

💥 Why You’ll Love It

  • ⚡ Unmatched Photorealism: Get ready for rich textures, natural lighting, and zero color-cast artifacts. Just use photography terms in your prompts (e.g., "50mm lens", "soft studio lighting") for stunning, professional-grade results.

  • 🎯 Reliable Brand Assets: Create marketing campaigns, product packaging, and UI mockups that actually feature your exact brand wording without embarrassing typos.

  • 🌍 Global Ready: With the new multilingual support, your Mates can easily generate localized visual content tailored for any international market.

  • ⏱️ Streamlined Workflows: Iterate quickly using the low quality setting to test concepts, then switch to high quality for the final masterpiece—all within the same conversation and with the same Mate.

🛠️ How to Get Started

  1. Navigate to Mates & Tools > Tool Store.

  2. Search for the OpenAI Image 2 tool and click + Create instance (it is ready to use immediately via our managed Pay-as-you-go billing, or you can add your own BYOK API key).

  3. Go to the configuration page of the Mate you want to empower, open the Tools tab, and click + Attach tool to add your new instance.

  4. Start creating! (Pro Tip: Put the exact text you want in quotes, and structure your prompts logically: Scene > Subject > Details > Constraints).

We can't wait to see the stunning visuals and innovative designs you'll bring to life with this state-of-the-art technology. Happy creating!

— The allmates.ai Team

April 27th, 2026

New

We are incredibly excited to announce the arrival of OpenAI's highly anticipated flagship model, GPT-5.5, on allmates.ai! The AI landscape is a relentless, high-speed race, and with this revolutionary release, OpenAI makes a resounding comeback, stepping up to confidently challenge Anthropic’s formidable Claude Opus 4.7.

But here is the best part: you don't have to choose a side. At allmates.ai, our core philosophy is simple—we believe in giving you the ultimate power of choice. You now have the world's most elite models at your fingertips, ready to empower your virtual team for any business challenge.

🔍 What’s New?

This update brings the absolute cutting-edge of generative AI directly into your collaborative workspace:

  • GPT-5.5 Integration: You can now equip your Mates with OpenAI's most advanced model to date, accessible directly through your existing API connections (OpenAI, Azure OpenAI, or OpenRouter).

  • The Ultimate Rivalry: GPT-5.5 redefines state-of-the-art performance, matching and in some areas surpassing the impressive benchmarks set by Claude Opus 4.7, particularly in complex logic, strategic analysis, and autonomous tool execution.

  • Unrestricted Ecosystem: We are expanding our agnostic model library. Your organization is never locked into a single provider. You can seamlessly pivot between OpenAI, Anthropic, Google, and Mistral depending on your immediate needs.

❤️ Why You’ll Love It

  • ⚡ The Power of Choice: Why settle for one ecosystem when you can have them all? You can now deploy a Mate powered by GPT-5.5 to conduct deep market research and data synthesis, while another Mate powered by Opus 4.7 reviews your financial reports and strategic memos. The freedom to choose the perfect engine for the job is your greatest competitive advantage.

  • 💥 Unprecedented Reasoning: GPT-5.5 brings a game-changing leap in logical deduction and multi-step problem-solving. It is designed to handle the most intricate enterprise workflows with fewer hallucinations and greater precision.

  • 🎯 Next-Level Agentic Performance: Build highly autonomous Mates. GPT-5.5's enhanced ability to understand deep context and execute sequential tool calls makes it the ultimate workhorse for complex, automated research and business intelligence.

  • ✨ Future-Proof Collaboration: By continuously integrating the latest market-leading models the moment they launch, we ensure your team is always equipped with the most powerful collaborative tools available.

🛠️ How to Get Started

Accessing GPT-5.5 is simple and takes just a few clicks:

  1. Navigate to your organization Settings and select Models Management.

  2. Activate GPT-5.5 via your preferred connection (OpenAI, Azure, or OpenRouter).

  3. Once activated, head to Manage Mates and assign GPT-5.5 to specific Mates, or set it as the default model for your entire organization.

We can't wait to see the incredible workflows you will build now that you have both GPT-5.5 and Claude Opus 4.7 at your disposal. Experience the ultimate freedom of choice today! Happy collaborating!

— The allmates.ai Team

April 21st, 2026

Improved

Anthropic not responding

Remember those frustrating error messages right in the middle of an important task?
They are now a thing of the past! Often caused by temporary overloads or outages at our AI providers, these interruptions are being replaced by rock-solid reliability thanks to our new Model Fallback feature.

🔍 What's New?

The Model Fallback equips your Mates with an automatic "Plan B." If your Mate's primary AI model experiences a failure (overload, outage, timeout), a backup model instantly takes over.

  • Invisible Continuity: The conversation continues naturally, with no red errors and no action required from you.

  • Total Transparency: When a message is generated by the backup model, a subtle amber icon appears in the message header. Simply hover over it to see exactly which model took over and why.

  • Administrator Control: By default, allmates.ai automatically selects the best backup model for you. Admins also have the flexibility to define a specific fallback model for the entire organization or disable the feature entirely.

❤️ Why You'll Love It

  • ⚡ Zero Frustration: Say goodbye to error messages that break your momentum. Your workflow is no longer dependent on external provider outages.

  • ⏱️ Time Saved: No more manually resending prompts or waiting for a provider to restore services. Your Mates will always deliver an answer.

  • 🛡️ Maximum Reliability: Enjoy a stable, fluid, and continuous experience. The feature is already active by default for all organizations (currently available for Native Mates).

🛠️ How to Configure It (For Admins)

Although the system works automatically out of the box, you retain full control over its behavior:

  1. Go to your Organization Settings.

  2. Navigate to Models Management > Settings.

  3. Under Default Model, use the Fallback Model dropdown to choose your preferred behavior (Automatic, a specific model, or None).

We are committed to providing you with the most stable and resilient AI collaboration experience on the market. Chat with peace of mind—your Mates always have your back!

— The allmates.ai Team

April 16th, 2026

New

The global tech community is buzzing with Anthropic's latest release: Claude Opus 4.7. As the AI landscape continues its rapid evolution, this new iteration delivers measurable improvements over Opus 4.6, specifically tailored for demanding enterprise workflows and complex knowledge work. The best news? You don't have to wait to experience it. Claude Opus 4.7 is available right now to join your virtual team on allmates.ai.

🔍 What’s New?

Claude Opus 4.7 shifts the focus toward deep reasoning, massive document analysis, and long-term memory.

  • 📄 Massive Document Processing: Opus 4.7 can now handle PDFs up to 600 pages long—a significant leap from the previous 100-page limit. Your Mates can now analyze comprehensive financial reports, extensive legal contracts, or entire research archives in a single prompt.

  • 🧠 Superior Long-Context Reasoning: Benchmarks show substantial improvements in how the model reasons over very large files. It doesn't just read the data; it effectively connects the dots across hundreds of pages to provide accurate, synthesized insights.

  • 💾 Enhanced Memory Retention: The model is now much better at utilizing file system-based memory. It retains important notes and context across long, multi-session projects, reducing the need to constantly remind it of past decisions.

  • 👁️ High-Resolution Vision: Opus 4.7 features improved visual acuity, processing images up to 3.75 megapixels. This makes it highly effective for extracting data from complex business diagrams, financial charts, or dense UI screenshots.

  • 🎯 Strict Instruction Following: The model now takes instructions much more literally. This precision ensures it executes your exact requirements for strict corporate workflows without skipping steps.

⚡ Why You’ll Love It

  • Built for the Enterprise: Opus 4.7 proves its worth in economically valuable knowledge work. It consistently outperforms competitors on independent evaluations like GDPval-AA, which measures performance across finance, legal, and business domains.

  • Consistent Pricing: You gain access to improved reasoning, better vision, and massive document handling at the exact same price point as Opus 4.6.

  • Immediate Integration: While the rest of the world is just reading the news, you can already assign Opus 4.7 to your Workspaces and Collabs.

🛠️ How to Activate It Right Now

Integrating the market's newest model into your workflow takes just a few clicks:

  1. Open your Mate's settings on allmates.ai.

  2. Select Claude Opus 4.7 from the list of available models.

  3. Start collaborating!

(Note: Because Opus 4.7 follows instructions more literally than its predecessor, you may want to slightly adjust your existing prompts for optimal results).

We are excited to provide you with immediate access to this new level of capability.

Happy collaborating, — The allmates.ai Team

April 14th, 2026

New

We are incredibly excited to announce a revolutionary update to allmates.ai! We have multiplied the number of available tools by five in our Tool Store. By integrating a massive array of new MCP (Model Context Protocol) servers, your Mates are taking an unprecedented leap forward.

They are no longer just assistants; they are becoming the central command hub for your entire software ecosystem. Whether you use HubSpot, Jira, Slack, GitHub, or dozens of other services, your Mates can now connect to them securely and interact with your data in real time.

🔍 What’s New?

  • A 5x Larger Tool Library: We've integrated a multitude of new MCP servers directly into the Tool Store. The possibilities for your Mates are now virtually limitless.

  • High-Security OAuth Connections: The vast majority of these new tools connect via the OAuth protocol. Say goodbye to managing complex API keys! A simple login window lets you authorize access in a standardized and highly secure way.

  • Total Personalization: Because authentication happens at the user level (via OAuth), the Mate acts on your behalf, with your exact access rights. It only sees what you are authorized to see, guaranteeing absolute confidentiality for your enterprise data.

  • Hundreds of Ready-to-Use Integrations: From Customer Relationship Management (CRM) to project management and team communication, connect your favorite apps in seconds.

❤️ Why You’ll Love It

  • Absolute Simplicity (One-Click Setup): Integrating a complex tool now takes just seconds. Click "Connect," validate access via your existing Microsoft, Google, or other accounts, and your Mate is ready to work.

  • 🛡️ Mastered Security & Governance: With OAuth authentication, you retain total control. You can revoke access at any time from the third-party application, and your data security is managed by the strictest industry standards.

  • 🎯 Infinite Automation Scenarios: Ask your "CRM Expert" Mate to retrieve the last 10 HubSpot transactions, analyze them, and then draft a follow-up email... all within the same conversation!

  • 🚀 A Future-Proof Platform: The mission of allmates.ai is clear: to provide the maximum number of integrations possible for your agents so you can automate all your business workflows. And this is just the beginning—many more tools are coming very soon!

🛠️ How to Get Started

Ready to supercharge your virtual team? It's incredibly simple:

  1. Navigate to Mates & Tools > Tool Store.

  2. Browse the new library and select the tool of your choice (e.g., HubSpot).

  3. Click Create instance, then use the authentication button to log in securely (OAuth) with your existing account.

  4. Go to the settings of your preferred Mate, open the Tools tab, and attach this new tool.

  5. Start a conversation! (e.g., "Can you check the latest ongoing transactions?")

We can't wait to see the incredible workflows and custom automations you will create with this new tool library. Connect your world, and let your Mates do the rest!

Happy collaborating! — The allmates.ai Team

April 1st, 2026

New

We are thrilled to announce a highly requested feature on allmates.ai: the ability to import and export your Mates! Whether you want to share your best AI coworkers across different organizations or simply create secure local backups, managing your Mates' lifecycle has never been easier.

🔍 What’s New?

Your Mates are now portable. With this update, you can:

  • One-Click Export: Download your Mate's complete configuration as a .zip file. This export includes its custom instructions, parameters, and all files attached to its Knowledge base.

  • Seamless Import: Create a new Mate instantly by uploading your .zip file into any workspace or organization.

  • Smart Reconfiguration: During the import process, the system allows you to select your preferred LLM model to ensure perfect compatibility with your new organization's connections.

⚡ Why You’ll Love It

  • Effortless Sharing: Built the perfect Data Analysis Mate? Easily transfer it to a client or another team working in a different allmates.ai organization.

  • Secure Backups: Create local copies of your complex configurations and knowledge bases for total peace of mind.

  • Standardized Deployment: Save valuable time by creating "template" Mates that you can duplicate and distribute across your company without ever having to rewrite prompts.

🛠️ How to Get Started

  1. To Export: Navigate to your Mate's configuration page, click the dropdown menu next to its name, and select Export. Your .zip file will be generated instantly.

  2. To Import: Go to the Mate creation menu, click on the new Import tab, and drop your .zip file.

  3. Finalize: Choose your preferred LLM model and validate!

💡 Important Note: For security and access management reasons (API keys), Tools are not included in the export. Don't forget to reattach your favorite tools (like Web Search or GitHub) to your Mate once it's imported!

We can't wait to see how you'll share and scale your AI workflows. Happy collaborating!

— The allmates.ai Team

March 31st, 2026

New

We are excited to introduce a simple yet incredibly powerful new feature to your workflow: Mate Duplication. You can now create an exact replica of any of your existing Mates with just a single click. Whether you want to safely experiment with new instructions or set up a robust cross-validation system, duplicating your Mates has never been easier!

🔍 What’s New?

  • One-Click Cloning: Instantly duplicate a Mate. The new "twin" automatically inherits all the original instructions, Knowledge base files, attached Tools, and parameters.

  • Independent Settings: Once duplicated, the new Mate is entirely independent. You can rename it, assign a different LLM model, or modify its permissions without affecting the original agent.

❤️ Why You’ll Love It

  • 🧪 Effortless Experimentation: Want to see how a different model handles your prompt? Duplicate your Mate, switch the model (e.g., from Gemini 3 Flash to Claude 3.5 Sonnet), and test the results side-by-side without breaking your original setup.

  • ✅ Advanced Cross-Validation: Combat hallucinations by using the "Twin Technique". Have your first Mate perform a complex data analysis, then ask your duplicated Mate (running on a different model) to double-check the work. It's an incredible way to ensure absolute accuracy.

  • ⏱️ Massive Time Savings: Stop rebuilding complex Mates from scratch. Use your best-performing Mates as templates to create specialized variations for different teams or tasks in seconds.

🛠️ How to Get Started

  1. Navigate to the Settings of the Mate you want to copy.

  2. Click the new Duplicate button.

  3. Rename your new Mate, adjust its parameters (like selecting a new LLM), and configure its access permissions. You're ready to collaborate!

We can't wait to see how you use this feature to build even more reliable and powerful AI teams!

— The allmates.ai Team

March 30th, 2026

Improved

We are excited to announce a highly requested quality-of-life improvement for managing your teams on allmates.ai! You now have complete flexibility over your collaborative environments with the ability to change a Workspace's privacy settings at any time.

🔍 What’s New?

Previously, a Workspace's visibility was set at creation. Now, your Workspaces can adapt dynamically to your project's lifecycle:

  • Dynamic Visibility: Switch any existing Workspace from Public to Private, or Private to Public with a simple click in the settings.

  • Retrospective Privacy: Close off a public workspace when a project becomes confidential. Once set to Private, only explicitly added members will retain access.

  • Open Up Silos: Transform a private, invite-only workspace into a public hub when you are ready to share your team's work, knowledge, and Mates with the entire organization.

❤️ Why You’ll Love It

  • ⚙️ Total Flexibility: Your workspace adapts to your needs, not the other way around. Evolve your access rights as your project grows.

  • 🔒 Enhanced Security: Instantly lock down a workspace to secure sensitive information or strategic discussions without needing to migrate your data to a new private room.

  • 🤝 Effortless Organization: Simplify how you manage team chats and collaborations. Easily open up successful private initiatives to the rest of the company to foster transparency.

🛠️ How to Get Started

Changing your workspace visibility is quick and easy:

  1. Navigate to your Workspace and click the ⚙️ Settings icon (Project settings & members).

  2. Go to the Details tab.

  3. Locate the Visibility dropdown menu and select either Public or Private.

  4. Click Confirm on the pop-up warning to apply your changes.

Log in now and take full control of your workspace privacy. Happy collaborating! — The allmates.ai Team