For anyone who has ever navigated a crowded room while hard-of-hearing or struggled through a conversation in a foreign city, the promise of live-captioning glasses has always felt like science fiction. The goal is simple: a heads-up display that turns spoken words into text in real-time, allowing the user to maintain eye contact and engage naturally without glancing down at a smartphone screen.
But as someone who spent years in software engineering before moving into reporting, I know that “simple” goals usually hide complex technical trade-offs. To get a clear, low-latency transcription on a pair of frames, manufacturers have to balance three conflicting variables: battery life, weight and processing power. If you want high accuracy, you usually need a cloud connection, which kills the battery and introduces lag. If you want offline privacy, you need a heavier processor and a bigger battery, which makes the glasses uncomfortable for long-term wear.
After extensive testing of the current 2026 market—including a deep dive into the latest hardware from Even, Leion, XRAI, AirCaps, and Captify—the industry is finally hitting a tipping point. We are seeing a shift away from bulky prototypes toward wearable tech that actually fits into a daily routine, though the “perfect” pair depends entirely on whether you value offline independence or financial predictability.
The Gold Standard for Value: Even G2
In a market defined by “subscription fatigue,” the Even G2 stands out as a disruptive outlier. Most of the competitors in this space treat the hardware as a gateway to a monthly recurring fee for transcription services. Even takes the opposite approach: there is no subscription plan. Everything is included out of the box.
From a user experience perspective, the G2 is the most seamless option for those who want a “set it and forget it” device. However, that simplicity comes with a technical constraint. The G2 is almost entirely dependent on an internet connection. Because it offloads the heavy lifting of speech-to-text processing to the cloud, it lacks meaningful offline features. If you lose your connection, the glasses effectively become expensive frames.
For the vast majority of users, Here’s a fair trade. The combination of high processing power and an affordable, one-time cost makes the G2 the most pragmatic choice for daily use, provided you have a reliable data plan.
The Hardware Twins: Leion and XRAI
One of the more intriguing discoveries in this testing cycle is the relationship between Leion and XRAI. While they market themselves as competitors, they share the same manufacturer. This is evident in the hardware: both weigh 50 grams without lenses (60 grams with them) and offer a similar battery life of roughly eight hours on a single charge, with a charging case providing significant additional juice.
Where these two diverge is in the software layer. Leion’s Hey 2 is currently the price leader, offering an intuitive app that includes a “free talk” two-way translation feature and a teleprompter—a tool that is surprisingly useful for public speakers or those with social anxiety. However, Leion’s pricing model is idiosyncratic. they sell “Pro minutes” rather than monthly subscriptions. For example, you can buy 120 minutes for $10 or 6,000 minutes for $200. This requires the user to manually toggle the Pro mode off to avoid draining their balance.
XRAI, by contrast, offers a more traditional subscription model ($20 to $40 per month) and supports a staggering 300 languages. While XRAI claims its display is brighter, the difference is negligible in real-world sunlight. The real advantage of XRAI is its rudimentary offline mode, which outperforms most of the competition, though its app interface is notably less user-friendly than Leion’s.
The Ergonomic Struggle: AirCaps and Captify
As we move toward the high-end and specialized segments of the market, the trade-offs become more pronounced. AirCaps offers one of the best offline experiences, but it fails the comfort test. At 53 grams without lenses, they are among the bulkiest frames tested. The battery life is also disappointing, lasting only two to four hours. To compensate, AirCaps sells “Power Capsules”—rechargeable clips that add 12 to 18 hours of life but add even more bulk to the frames.
Then there is Captify, the luxury option that can cost up to $1,399 with prescription lenses. On paper, Captify is the most svelte, weighing only 40 grams. But luxury doesn’t always equal performance. In testing, the prescription lenses provided by Captify were the blurriest of the group, which is a critical failure for a device that requires the user to read little text in their line of sight. While it supports offline transcription, the performance drops significantly without internet, and translation fails entirely.
| Model | Weight (no lens) | Battery Life | Pricing Model | Offline Support |
|---|---|---|---|---|
| Even G2 | Competitive | High | One-time purchase | Poor |
| Leion Hey 2 | 50g | 6–8 Hours | Per-minute credits | None |
| XRAI | 50g | Up to 8 Hours | Monthly/Minute | Basic |
| AirCaps | 53g | 2–4 Hours | Monthly | Solid |
| Captify | 40g | 4 Hours | Monthly | Limited |
The Technical Divide: Cloud vs. Edge Computing
The disparity in these devices highlights a broader tension in AI hardware: the battle between cloud-based APIs and “edge” computing (processing data on the device itself). Devices like the Even G2 rely on the cloud, which allows them to remain lightweight and affordable but leaves them useless in “dead zones.”
AirCaps and XRAI are attempting to bring more processing to the edge. This is essential for users who prioritize privacy or those who travel to areas with spotty connectivity. However, until we see a breakthrough in low-power NPU (Neural Processing Unit) integration for eyewear, users will continue to choose between a comfortable frame with a subscription or a bulky frame that works offline.
For those seeking a recommendation, the choice is clear: if you have consistent internet access, the Even G2 is the most economical and powerful tool. If you require offline capabilities and can tolerate a heavier frame, AirCaps is the viable, albeit clunky, alternative.
The next major milestone for this category will be the integration of more sophisticated on-device LLMs, which should theoretically allow for high-accuracy translation without the need for a cloud handshake. We expect the first wave of these “True Edge” captioning glasses to enter the certification process by late 2026.
Do you use live-captioning tech or assistive hearing devices? Share your experience in the comments or let us know which feature you value most in wearable AI.
