The Biggest ARMES Update Yet: New ARMES GPT, 6 New Models, Command View & More

This one is big. Genuinely, the most significant update we've ever shipped. A brand new ARMES GPT across every tier, six new frontier models, a brand new AI workspace, a completely modernized platform, and so much more.

If you read nothing else: sign out of ARMES and sign back in. That's it. You'll get the best version of ARMES we've ever shipped. There's a lot to share, and every section has something meaningful for you. Here's what's inside:

1. ARMES GPT: Rebuilt from the Ground Up
2. The Frontier Model Explosion: 6 New Models
3. Command View: Your New AI Workspace
4. Under the Hood: Platform Improvements
5. iOS Beta & Onboarding
6. Action Required: Clean Update
7. Updated Privacy Policy & Terms
8. The Road to Launch

---

Every tier of ARMES GPT has received a major upgrade: Free, Pro, and Ultra. This isn't a patch. It's a rebuild. ARMES GPT is now more powerful, more aware of the ARMES ecosystem, and more capable than any version before it.

Agent-to-Agent Collaboration (Pro + Ultra)

ARMES GPT can now message and collaborate with other ARMES agents directly. Ask it to coordinate with your specialized agents, delegate tasks, and bring back results, all within a single conversation.

Your Knowledge Base, Managed by AI (All Tiers)

ARMES GPT can now create, edit, and update your notes and context directly. Ask it to save a summary, update a document, or master plan your knowledge base. It just does it. Your private context, managed intelligently.

Deeply Centered in Your Ecosystem (All Tiers)

ARMES GPT now understands where it sits in your ARMES ecosystem. It knows about your agents, your notes, your tools. It can guide you, help you build, and make the most of everything ARMES offers. This is a modernization designed to compete with the best AI experiences in the world. We believe it does.

---

The last two weeks have been one of the most significant periods in AI model releases we've ever seen. Six new frontier models from five different AI labs have shipped in rapid succession. Every single one of them is now live on ARMES.

• Claude Opus 4.6 (Ultra) — Anthropic, Feb 5. Anthropic's most powerful model ever. Deep reasoning, creative writing, complex coding, and strategic thinking at the highest level.

• Claude Sonnet 4.6 (Pro + Ultra) — Anthropic, Feb 17. The surprise of the month. Matches or beats the flagship Opus on real-world knowledge work and leads all models on financial analysis. Developers preferred it 70% of the time over its predecessor.

• Gemini 3.1 Pro (Pro + Ultra) — Google, Feb 19. Google's answer to everything. #1 on abstract reasoning benchmarks, competitive coding, and scientific reasoning. Leads 13 of 16 evaluated benchmarks. A genuine leap.

• GLM-5 (All Tiers) — Z.ai, Feb 11. Z.ai's flagship open-weight model. State-of-the-art agentic engineering, complex coding, and long-horizon reasoning. Top-ranked among open-source models on operational benchmarks.

• KIMI K2.5 (All Tiers) — Moonshot, Jan 27. Moonshot's agentic powerhouse. Advanced tool use, strong mathematical reasoning, and a unique agent swarm system that can orchestrate up to 100 sub-agents in parallel. Leads on agentic benchmarks.

• MiniMax M2.5 (All Tiers) — MiniMax, Feb 12. MiniMax's productivity powerhouse. Matches the best on real-world coding benchmarks, excels at agentic search and tool use, and handles full development lifecycles across 10+ languages. Open-weight.

Here's the important context: there is no single best AI model anymore. Each model has become deeply specialized. A user asking for help with a tax strategy should get a different model than one debugging a React component or writing a short story. Picking the right model for the right task now matters more than ever.

This is exactly where ARMES was built to operate.

Auto Mode: Refreshed

We've completely refreshed the Auto Mode intelligence layer to incorporate the latest benchmark data and real-world performance from these new releases:

• Finance question? Auto Mode routes to Sonnet 4.6, the #1 model for financial reasoning.
• Complex coding architecture? Opus 4.6, the flagship for high-stakes development.
• Abstract logic? Gemini 3.1 Pro, the reasoning champion.
• Calculus problem? GPT-5.2, still unmatched in pure mathematics.
• Quick question? Haiku 4.5. Instant, smart, no wasted resources.

You shouldn't have to be an AI expert to get expert-level AI. That's what Auto Mode is for. The right model, for the right task, every time. It really is a beautiful thing seeing ARMES GPT utilize the intelligence of multiple frontier AI labs across your tasks and needs.

Tier Breakdown

Free	Pro	Ultra
GLM-5, KIMI K2.5, MiniMax M2.5, and more	Everything in Free + Sonnet 4.6, Gemini 3.1 Pro, GPT-5.2, and more	Everything in Pro + Opus 4.6, the premium powerhouse for moments where the absolute highest quality matters

Privacy Across Every Model

Every single model on ARMES runs with private inference. Every lab. Every tier. No exceptions. The providers process your request and immediately incinerate it. No retention, no profiling, no monetization. Your intelligence stays yours.

---

Working with multiple AIs, multiple models, and various contexts has never been easier. We're introducing a brand new unified workspace for your entire ARMES ecosystem.

Inspired by agentic code workflows and tools like VS Code and Cursor, Command View lets you have multiple chats with different models or agents and multiple notes open simultaneously, all in a tabbed experience.

• Multiple concurrent chats with different models and agents
• Notes and context open side-by-side with your conversations
• Tabbed interface: organize and switch between sessions seamlessly
• Manage your ARMES settings and individual agent configurations in tabs alongside your work
• Works on mobile and desktop. Your entire ARMES experience in one dedicated view. Chat, context, settings, agents. All of it.

---

We've modernized and improved our total platform architecture. You'll feel it the moment you open ARMES. Faster load times, snappier navigation, and a more capable foundation across every surface.

• Attachments & File Import — Completely hardened. More reliable, more capable, and significantly faster. Upload with confidence.
• Chat & Context Pages Refined — Special attention paid to improving the dedicated Chat and Context workspaces. These work beautifully on both desktop and mobile, and stay in sync with Command View.
• Automatic Chat Naming — New chats with any agent are automatically given descriptive names. Many of you have already noticed and praised this one. Thank you.
• Note & Folder Archiving — Archive notes and note folders to keep your active knowledge base focused and current. Archived content is still viewable and editable. In a future update, archived notes will be siloed from AI inference entirely, giving you full control over what your AI can and can't access.
• Prompt Templates with Pinned Context — Prompt templates now support having specific notes or context already attached. When you add a template to any chat, the prompt is inserted ready to send or edit, and any pinned notes come along automatically.
• In-Chat Commands — A reminder: from any chat surface in ARMES, type @ to tag specific notes or context for your query, and / to see available commands.

---

The iOS beta has been going incredibly well. As we put the finishing touches on this major ecosystem update, we'll likely push one more update to the iOS beta. We are so close to being ready for the official App Store launch.

We've also taken extra care with the onboarding experience for new ARMES users. Command View has a guided tour, and we've added a variety of guided experiences across the platform to help users understand everything their ecosystem can do.

Not in the iOS beta yet? Join via TestFlight.

---

Due to the depth and scope of this update, a very small group of users may see unexpected behavior in their app. To ensure you receive the cleanest possible experience with the brand new ARMES, we're recommending one simple step:

Sign out of ARMES on every surface you use. Then sign back in.

This ensures a clean implementation of the entire update. It takes 30 seconds and guarantees you're running the latest ARMES.

---

As part of this update, we've comprehensively rewritten and strengthened both our Privacy Policy and Terms of Service. We rewrote them from scratch because you deserve to know exactly how your data is handled. Plain language. No ambiguity. No legalese designed to confuse.

What we've made clearer than ever:

• We do not sell your personal data
• We do not use your data for advertising
• We do not build profiles on you
• AI providers cannot learn from you, profile you, or monetize your data. It is architecturally impossible for them to retain it
• You own your input and your output
• We do not use your content to train AI models

Read the Privacy Policy → Read the Terms of Service →

---

We are getting our ducks in a row. Everything in this update is building toward launch. The rebuilt ARMES GPT, the new models, Command View, the platform hardening, the legal refresh. All of it. We'll be sharing the exact launch date and details in the coming weeks.

When the moment comes, we'd love for the community to rally behind ARMES and really give this platform a fair shot at greatness. Here's what will make a difference:

• Share Your Referral Code — Invite friends. You both get $10. Find your code in Settings → Referrals.
• Rate & Review on the App Store — Once we launch, an honest review goes a long way for visibility.
• Support Our Product Hunt Launch — We'll share the link when the time comes. An upvote and comment would mean the world.
• Share on Social Media — Tell your network. Share your experience. Tag @armesai. Word of mouth is everything.

Pricing: Lock In the Beta Rate

ARMES Pro still has a 10-day free trial. The beta price is $19/month for access to all top AI with private inference. After our official launch, it returns to the standard price of $25/month.

Beta Price (Now)	After Launch
$19/month	$25/month

Current subscribers: You're locked in at $19/month. This beta rate is still available for anyone who subscribes before launch.

---

I use ARMES as my main daily driver for AI. Every day. I'm working to make it the best experience possible, and hearing from users is the most valuable thing in the world right now.

We're hungry for feedback. We're hungry for testimonials. So many of you have already reached out and shared your thoughts, your requests, and your kind words. It's been wonderful and invaluable.

If you'd like to share feedback, request a feature, or even jump on a quick Zoom to talk about AI and how you're using ARMES, I'd genuinely love that. Reach out anytime: joseph@armes.ai

The frontier isn't one model anymore. It's an ecosystem. ARMES is the only platform that gives you the entire ecosystem working together, intelligently, under your control. And with private inference, none of it ever leaves your hands.

Thank you for being part of this. We're building something special here, and every one of you is a part of it.

Private inference of frontier intelligence.

Joseph Founder, ARMES