Meta’s Two New Smart Glasses Are Slimmer and More Like Regular Glasses
New models have prescriptions in mind for people who actually still need help seeing.
Meta has two new pairs of AI smart glasses that are geared toward a thing that glasses have been doing for quite a while—helping you see. The Ray-Ban Meta Blayzer Optics and Scriber Optics are more focused on people who wear prescription lenses and also come in new styles that differentiate them from existing Ray-Ban-branded smart glasses.
While the Blayzer is what Meta calls a “sleek rectangular frame” and comes in standard and large sizes, the Scriber has a more rounded frame. These frames are slimmer than Meta’s existing AI glasses styles such as the Wayfarer, Headliner, and Skyler. According to Meta, both pairs “support nearly all prescriptions” and come with changeable nose pads and adjustable temple tips. Styles include matte black, transparent black, and transparent dark olive, but also come in “seasonal colors” like transparent “Matte Ice Grey” and transparent “Stone Beige.” One thing that’s a little bit different from the previous Ray-Ban lineup is that the carrying/charging case is a dark brown instead of the light brown of previous generations.
For the record, neither of these glasses has a screen in them like the Meta Ray-Ban Display—these are based on Meta’s non-display AI glasses.
© Meta
© Meta
At 49g, both standard-sized pairs of the new smart glasses weigh about the same as the standard-size Ray-Ban Meta Gen 2 AI glasses, which have a weight of 51g. The large size weighs 52g.
Outside of a different look and more prescription lens compatibility, Meta’s new Blayzer Optics and Scriber Optics are basically the same AI glasses that the company has been selling, with the same hardware and features, including a 12-megapixel camera, speakers, 8 hours of battery, built-in Meta AI, and computer vision. Whether Meta plans to expand features to cater more to people with lower vision is anyone’s guess, but they’re more poised to do so now.
Both pairs of smart glasses are available for preorder starting today from Meta.com and Meta retail partners in the U.S., as well as “select international markets” starting April 14. Pricing for both pairs starts at $499.
Sign in to highlight and annotate this article

Conversation starters
Daily AI Digest
Get the top 5 AI stories delivered to your inbox every morning.
More about
modelnew model
Unmasking the AI Crypto Trading Revolution: Real Opportunities vs. Digital Deception in 2024
<h1> Unmasking the AI Crypto Trading Revolution: Real Opportunities vs. Digital Deception in 2024 </h1> <p><a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fevebbp0g3w96qrnigp5h.jpeg" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fevebbp0g3w96qrnigp5h.jpeg" alt="Blog Image" width="800" height="649"></a></p> <p>The promise of effortless wealth through AI-driven crypto trading platforms like Instant Avita Ark is captivating, but the reality often hides a complex landscape of innovation and outright

9 MCP Production Patterns That Actually Scale Multi-Agent Systems (2026)
<h1> 9 MCP Production Patterns That Actually Scale Multi-Agent Systems (2026) </h1> <p>Model Context Protocol went from "interesting spec" to industry standard in under a year. 97 million monthly SDK downloads. Every major AI provider on board — Anthropic, OpenAI, Google, Microsoft, Amazon.</p> <p>But most tutorials still show toy examples. A weather tool. A calculator. Cool for demos, useless for production.</p> <p>Here are 9 patterns we've battle-tested in real multi-agent systems — with code you can ship today.</p> <h2> 1. The Tool Registry Pattern </h2> <p>Don't hardcode tools. Register them dynamically so agents discover capabilities at runtime.<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight typescript"><code><span class="c1">// mcp-registry/src/registry.ts</

8 AI Agent Memory Patterns for Production Systems (Beyond Basic RAG)
<h1> 8 AI Agent Memory Patterns for Production Systems (Beyond Basic RAG) </h1> <p>Every AI agent tutorial shows stateless request-response. User asks, agent answers, context vanishes.</p> <p>Real agents need memory. Not just "stuff the last 10 messages into the prompt" — actual structured memory that persists, compresses, and retrieves intelligently.</p> <p>Here are 8 memory patterns we use in production, ranked from simplest to most sophisticated.</p> <h2> 1. Sliding Window with Smart Summarization </h2> <p>The baseline. Keep recent messages, summarize old ones. But do it properly.<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight python"><code><span class="c1"># memory/sliding_window.py </span><span class="kn">from</span> <span class="n">dataclasses</span> <span c
Knowledge Map
Connected Articles — Knowledge Graph
This article is connected to other articles through shared AI topics and tags.
More in Models
ciflow/torchtitan/178948: [dynamo] Implement tp_as_number->nb_or slot
<p>Replace the ad-hoc isinstance cascades in call_or_/call_ior with a generic<br> binary-op dispatch that mirrors CPython's binary_op1 algorithm. Each VT<br> subclass that supports | now provides an nb_or_impl (and nb_ior_impl for |=)<br> slot method, following the pattern established by nb_index_impl and nb_int_impl.</p> <p>The generic infrastructure in object_protocol.py (binary_op1, binary_op,<br> binary_iop) handles forward/reverse dispatch and NotImplemented propagation.<br> Per-type slots are implemented on ConstantVariable (int/bool/frozenset/type),<br> SetVariable, ConstDictVariable, DictKeysVariable, SymNodeVariable,<br> UserDefinedClassVariable, UserDefinedObjectVariable, UserDefinedDictVariable,<br> UserDefinedSetVariable, and OpaqueObjectClassVariable.</p> <p>Authored with Clau
trunk/d52b2f548aa3cfcfcd499fcba764fccf29628de6: [inductor] Enable precompiled headers in fbcode (#178870) (#178870)
<p>Summary:</p> <p>Enable cpp_cache_precompile_headers for fbcode</p> <p>Authored with Claude.</p> <p>Test Plan: CI</p> <p>Reviewed By: IvanKobzarev</p> <p>Differential Revision: D98767213</p> <p>Pull Request resolved: <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4179300248" data-permission-text="Title is private" data-url="https://github.com/pytorch/pytorch/issues/178870" data-hovercard-type="pull_request" data-hovercard-url="/pytorch/pytorch/pull/178870/hovercard" href="https://github.com/pytorch/pytorch/pull/178870">#178870</a><br> Approved by: <a href="https://github.com/IvanKobzarev">https://github.com/IvanKobzarev</a></p>

Discussion
Sign in to join the discussion
No comments yet — be the first to share your thoughts!