Woofun AI reports that the Leikeji Thunderbird V4 has fundamentally shifted the competitive landscape for AI eyewear by resolving long-standing hardware deficiencies, specifically battery capacity and wearability, while simultaneously exposing that the core limitation of the category now resides in the maturity of the underlying artificial intelligence logic rather than physical constraints. The device demonstrates that while the form factor has reached a point of practical utility with a 38-gram base weight and a 249.6mAh battery, the software layer remains inconsistent, delivering accurate real-time data in some instances while failing to recognize established market realities in others. This divergence suggests that the industry has outpaced its own software development cycles, creating a scenario where the hardware is ready for daily deployment but the intelligence driving it is not. The review of the Thunderbird V4 reveals a product that has successfully addressed the physical friction points of previous generations, such as the V3, but has not yet solved the cognitive friction inherent in current AI agent interactions.

The evolution of AI glasses over the past year mirrors the chaotic proliferation phase of smartphones more than a decade ago, characterized by monthly model releases and a unified manufacturer narrative that positions these devices as the inevitable future of human-computer interaction. The theoretical logic supporting this trajectory is robust: glasses are naturally positioned near the eyes and ears, equipped with cameras, microphones, and speakers to detect the environment without requiring the user to extract a separate device like a smartphone.

However, the transition from theoretical potential to practical utility has been hindered by fundamental physical barriers, including continuous wear comfort, battery endurance, and audio fidelity. If a device fails to meet these basic ergonomic thresholds, no amount of advanced AI capability can justify regular usage. The Thunderbird V4 represents a critical inflection point where these physical barriers have been largely dismantled, allowing the true nature of the remaining challenges to surface. The device weighs 38 grams at factory release, a reduction from the previous model's 39 grams, and reaches 49 grams when fitted with myopic lenses exceeding 400 degrees with a 1.60 refractive index. For context, standard myopic glasses weigh approximately 24 grams. While the weight differential is noticeable, the wearing experience for four to five hours remains comfortable for long-term users, though pressure on the temples becomes perceptible after eight to nine hours of continuous use. The design incorporates air-filled nose pads to distribute pressure more evenly than traditional hard pads, mitigating discomfort to a level comparable to regular eyewear. These subtle engineering choices are as critical as the raw weight reduction, as they address the psychological and physical fatigue associated with all-day wear. The aesthetic design successfully mimics traditional glasses, avoiding the conspicuous look of earlier prototypes, a result achieved through rigorous restraint in the product design process regarding every gram and inch of the frame structure. Despite these successes, minor mechanical imperfections remain, such as the damping effect when opening and closing the temple frames, which lacks the smoothness expected at this price point, though it does not impede normal function.

A frequently overlooked but critical specification of the Thunderbird V4 is its IP67 dust and water resistance rating, which significantly alters the user's psychological relationship with the device in dynamic environments. This feature allows users to operate the glasses while sweating, in light rain, or in complex surroundings without the constant anxiety of water damage, a concern that previously limited the utility of similar devices. For a product intended for regular, all-day wear, eliminating this psychological barrier is a vital component of the overall user experience. The open-air audio architecture presents a distinct set of trade-offs that define the listening experience. In quiet environments, the Thunderbird V4 delivers decent sound quality with a sense of depth and sufficient bass for music and podcasts.

However, in noisy settings such as subways, shopping malls, or roadside locations, the lack of sound isolation becomes a double-edged sword. While it ensures safety by allowing ambient noise to penetrate, the same ambient noise can easily mask the audio output from the speakers. Compared to earclip-style TWS headphones like the Huawei FreeClip or earhook-style models like the Ola Friend, the sound-emitting units of the glasses are positioned farther from the ear canal, which degrades audio clarity in high-noise scenarios. To mitigate this, the full-set package includes sound-guiding fins designed to reduce sound leakage and improve clarity, offering a partial solution to the inherent limitations of the open-ear form factor. The dual indicator light system, featuring both an internal and external light, represents a thoughtful design evolution from the V3 model. While many competitors rely solely on an external light to alert bystanders that the camera is active, the Thunderbird V4 includes an internal indicator facing the wearer. Initially perceived as superfluous, this internal light proves essential in noisy environments where visual confirmation is necessary, particularly indoors during the day or at night when the light may be dazzling but the accompanying warning sound provides clear status updates regarding photo or video modes.

The imaging capabilities of the Thunderbird V4 are anchored by a significant hardware upgrade in the left-side camera, which serves as the primary photography tool. The device utilizes a Sony IMX586 sensor with a 1/2.9-inch size, a resolution of 9 million pixels, and a pixel size of 2.09μm². This sensor is paired with a 5P lens featuring an equivalent focal length of 17 mm and an f/2.2 aperture, enabling high-quality 2.5K video recording. This specification set is notably high compared to other AI glasses on the market. A unique aspect of the sensor design is its 1:1 square shape, a strategic choice necessitated by the fixed nature of glasses on the face, which prevents the rotation required to switch between landscape and portrait modes on smartphones. The square sensor captures a more complete image before cropping, theoretically reducing the loss of visual information and image quality associated with format conversion.

However, a significant software limitation currently prevents users from exporting photos in the original 1:1 ratio; users are restricted to 4:3 or 3:4 formats and must manually crop them to square, thereby limiting the full potential of the sensor's flexibility. In terms of actual image quality, a noticeable gap remains between the Thunderbird V4 and high-end smartphones, a disparity driven by differences in lens size, sensor specifications, and processing power. It is unrealistic to expect glasses to compete directly with smartphones in raw imaging quality, yet for everyday photography, the device is more than sufficient. The ultra-wide-angle lens enhances the sense of presence in first-person shots, capturing a broader field of view, while the software effectively controls edge distortion common in wide-angle lenses. The automatic horizon correction feature in the editing software ensures that most photos are usable directly, addressing the common issue of tilted horizons caused by natural head movements during shooting. The color rendering is realistic, avoiding the oversaturation and artificial enhancement often seen in smartphone cameras, resulting in images that better reflect the authentic purpose of first-person photography. Video recording benefits from robust anti-shake performance, maintaining stability during activities like walking. In well-lit daytime environments, the images retain detailed environmental information and proper spatial layers with acceptable smearing, though purple fringes may occasionally appear in high-contrast areas like tree branches and building edges without significantly impacting the viewing experience.

The interactive workflow of the Thunderbird V4 follows the industry standard of using buttons and voice commands for photography and video recording, with short presses triggering photos and long presses initiating video. The process is straightforward, but the post-capture workflow reveals critical differentiators in user experience. The mobile app automatically applies horizon correction during editing, a feature that addresses the inability of AI glasses to offer real-time preview functionality. Since users cannot manually adjust composition through a viewfinder, they must rely on rough directional estimation during shooting, making the automatic correction essential for improving image quality and reducing friction. The app also provides comprehensive editing tools for adjusting image formats and details, contributing to a smooth photography experience that leverages the larger square sensor and refined pre-processing steps. The most impressive aspect of the Thunderbird V4 is its battery life, which has been transformed through significant engineering changes. In a test scenario, the device was activated at 11 a.m. with 95% battery and dropped to 46% by 9:20 a.m. the following day, lasting nearly half a day despite intermittent music listening and audio recording. Even with increased usage of the AI voice assistant and photography functions in subsequent days, the battery life remained comparable to a smartphone, requiring only one charge per day. This represents a dramatic improvement over the Thunderbird V3, which suffered from insufficient battery life that could not sustain several hours of music playback. The achievement is attributed to the adoption of silicon negative electrode battery technology and steel shell packaging similar to that used in the iPhone, increasing the battery capacity by 57% from 159mAh in the V3 to approximately 249.6mAh in the V4. Combined with a dual-core architecture and software optimizations, this hardware upgrade has resolved a critical bottleneck for the category. The charging speed further enhances usability, with the device capable of charging from 5% to 100% in less than half an hour, allowing users to recharge during routine activities like showering or eating without disrupting their daily schedule.

However, the interaction interface of the Thunderbird app requires improvement, as repeated swiping gestures are needed to return to the main menu, introducing unnecessary complexity that contradicts the goal of natural and quick interactions.

The AI experience provided by the Thunderbird V4, running on vAROS 2.2.38, remains the most contentious aspect of the device. Lei Hongwei, the CEO of Leikeji Innovation, stated at the launch event that the focus for this generation would shift from developing new features to optimizing the basic user experience. In practice, the voice assistant 'Xiaolei' handles the vast majority of interactions, while camera buttons are dedicated to media capture and touch controls manage volume and song switching. Despite the multi-microphone array enabling accurate voice recognition in noisy environments, the response latency of the voice command system is inconsistent. Instances of the assistant failing to respond to commands, such as playing the news radio, until multiple attempts are made, have led users to prefer touch controls. This inconsistency points to underlying issues with the voice recognition mechanism.

Furthermore, the AI system exhibits a troubling duality in its data accuracy. While it can sometimes provide precise information such as announcing the release date of GTA 6 on June 25th, it frequently relies on outdated information, claiming that the Switch 2 has not been released despite it being available for a year. This discrepancy highlights the limitations of the current AI models integrated into the device. Additional functions, including payment, AI-powered recording, note-taking, news radio access, English learning tools, translation services, and currency conversion, are available through the AI Agent.

However, these functions operate more like simple 'robots' executing predefined tasks rather than advanced systems like OpenClaw. The content and voice of the news radio function clearly exhibit signs of AI generation, with a quality that, while not inherently bad, lacks the polish required for repeated engagement. The advantages of the Thunderbird V4 are clear: significantly improved battery life meeting all-day needs and an enhanced wearing experience closer to traditional glasses. The disadvantages are equally distinct: an unsmooth mobile app interaction interface and a voice recognition function that requires further refinement. Ultimately, the Thunderbird V4 demonstrates that the hardware evolution of AI glasses has reached a mature stage, but the application of AI technologies remains the primary bottleneck. The strategy of pausing new feature development to optimize the basic user experience is validated by the device's performance, yet the rapid development of large-scale AI models suggests that the software layer must catch up to the hardware to fully realize the potential of this form factor. The real challenge for the industry is no longer about adding more AI functions but about how these technologies are effectively utilized to enhance the user experience.