iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
An October 1, 2026 DEV Community article by Zakaria Khchiche reports three bugs in OGX, an open-source AI agent framework: an Elasticsearch hybrid-search setting that was silently ignored, pgvector list filters that mishandled booleans and numbers, and token-detail counters that failed to accumulate across model calls. The author described OGX as a continuation of Meta’s Llama Stack project, but that description and the reported bugs have not been independently verified here. The findings should therefore be read as the author’s account, not as confirmation of current behavior or fix status.
What the reported bugs have in common
Each defect could produce a plausible outcome without an obvious error: a search could run with a different setting than requested, a metadata filter could return the wrong documents, or usage details could undercount activity. As the article’s author put it, “A bug that crashes gets noticed right away. A bug that returns a plausible but wrong result can run in production for months without anyone noticing.”
The practical lesson is to verify outcomes, not just successful calls: inspect the configuration passed to the search backend, test expected document sets for filters, and check detailed usage totals across chained calls.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallElasticsearch hybrid search: a setting was reportedly dropped
The article says the allowed RRF parameter list used the misspelled name rank_windows_size, with an extra “s.” As reported, a user supplying the correct rank_window_size was treated as providing an unknown setting, so the value was silently discarded.
#1 Best Overall
![Arduino® UNO™ Q 4GB [ABX00173]- Hybrid Board, Qualcomm Dragonwing QRB2210 microprocessor (MPU) & STM32U585 Microcontroller(MCU), AI Vision, Voice, IoT, Robotics, Linux Debian OS, Wi-Fi 5, USB-C](https://m.media-amazon.com/images/I/51cl4xygwGL._SL160_.jpg)
- Dual-Brain Hybrid Power: Combines the Qualcomm Dragonwing QRB2210 MPU (Quad-core Arm Cortex-A53 @ 2.0 GHz CPU, Adreno GPU, AI acceleration) and the real-time, low-power STM32U585 MCU for advanced applications like object recognition, voice commands, and motion detection.
- AI & Linux Capabilities: Unlocks AI-powered vision and sound solutions; runs Linux Debian OS for coding in Python and supports the Arduino ecosystem with libraries and Sketches; quick start with Arduino App Lab.
- Advanced Features: Equipped with 4 GB LPDDR4 RAM, 32 GB eMMC built-in storage, ideal for single-board computer (SBC) mode, running multiple simultaneous high-level processes, more complex AI or ML models, extensive logs. Dual-band Wi-Fi 5 (2.4/5 GHz), Bluetooth 5.1, and high-speed headers for vision, audio, and display peripherals.
- Seamless Expansion & Connectivity: Features the classic UNO form factor for shields compatibility, an 8x13 LED matrix, and a Qwiic connector for easy expansion with Modulino nodes; power and connect via the USB-C connector.
- Intended Use & Development: The perfect platform for prototyping robotics or IoT projects, empowering innovators with a unified development experience to mix Arduino Sketches, Python scripts, and containerized AI models in a single interface.
The example requested a window size of 50, while the prior behavior used 10. A successful request alone would not show whether the requested value was applied; the meaningful check is the setting actually passed to Elasticsearch. The author said a correction was submitted as pull request #6697. The reviewed article does not establish whether it was merged or included in a release.
pgvector filters: boolean and numeric values reportedly mismatched
The author reports that IN and NOT IN filters converted values to Python text before comparison. For a boolean, Python True becomes "True", while PostgreSQL’s JSON metadata extraction yielded "true" in the described case. The article also describes a numeric mismatch between 2023.0 and 2023.
Rank #2
- Dual-Brain Hybrid Power: Combines the Qualcomm Dragonwing QRB2210 MPU (Quad-core Arm Cortex-A53 @ 2.0 GHz CPU, Adreno GPU, AI acceleration) and the real-time, low-power STM32U585 MCU for advanced applications like object recognition, voice commands, and motion detection.
- AI & Linux Capabilities: Unlocks AI-powered vision and sound solutions; runs Linux Debian OS for coding in Python and supports the Arduino ecosystem with libraries and Sketches; quick start with Arduino App Lab.
- Advanced Features: Equipped with 2 GB LPDDR4 RAM, 16 GB eMMC built-in storage, ideal to develop in PC-connected mode, running the OS, Python scripts, and basic network services (SSH) without a demanding GUI or heavy multitasking; great for lightweight AI and memory-optimized TinyML applications, needing local storage for basic OS and core libraries. Dual-band Wi-Fi 5 (2.4/5 GHz), Bluetooth 5.1, and high-speed headers for vision, audio, and display peripherals.
- Seamless Expansion & Connectivity: Features the classic UNO form factor for shields compatibility, an 8x13 LED matrix, and a Qwiic connector for easy expansion with Modulino nodes; power and connect via the USB-C connector.
- Intended Use & Development: The perfect platform for prototyping robotics or IoT projects, empowering innovators with a unified development experience to mix Arduino Sketches, Python scripts, and containerized AI models in a single interface.
The risk is especially clear with NOT IN: if a value that should match an exclusion is missed, a document can remain in the results. In a RAG workflow, that could allow unpublished or outdated material into context. The author reported testing on PostgreSQL 16 with two documents; these are the article’s test conditions, not an independent reproduction.
| Reported filter | Before the reported fix | After the reported fix |
|---|---|---|
published in [true] |
No result | Document 0 |
published not in [true] |
Documents 0 and 1 | Document 1 |
year in [2023] |
No result | Document 1 |
The author said the fix was submitted in pull request #6698. Its merge and release status are not established by the reviewed article.
Rank #3
- Single core ARM Cortex-A7 32-bit core, integrated with NEON and FPU
- Built in Micro's self-developed 4th generation NPU, with high computational accuracy and support for mixed quantization of int4, int8, and int16. Among them, int8 has a computing power of 0.5 TOPS and int4 has a computing power of up to 1.0 TOPS
- Built in self-developed 3rd generation ISP3.2, supports 4 million pixels, and supports various image enhancement and correction algorithms such as HDR, WDR, and multi-level denoisin
- It has powerful encoding performance, supports intelligent encoding, adapts to save bit rates according to the scene, and saves more than 50% of the bit rate compared to conventional CBR mode, making the captured images high-definition, smaller in size, and doubling the storage space
- The design with built-in RISC-V MCU supports low-power fast startup, 250ms fast capture, and simultaneous loading of AI model library, enabling facial recognition to be completed within 1 second
Usage details: token counters reportedly kept only the last call
According to the article, total input and output tokens were summed across multiple model calls, but cached-token and reasoning-token details were overwritten by the final call rather than accumulated. In the example, cached-token values of 80 and 120 should total 200, but the reported pre-fix detail was 120. Reasoning-token values of 10 and 15 should total 25, while the reported pre-fix detail was 15.
The author noted that these details feed cost tracking: an understated cached-token count understates caching, and an understated reasoning-token count understates reasoning usage. The article identifies issue #6699 and proposed commit e8177d0; it does not establish that a fix shipped.
Quick Recap
Rank #4
- 【POWERFUL ESP32‑S3 CONTROLLER】Built‑in Xtensa 32‑bit LX7 dual‑core processor, 512KB SRAM, 8MB PSRAM, 16MB Flash for stable AI voice computing and multitask processing.
- 【Preloaded Dual AI Platforms】Comespre-installed with complete Deepseek and OpenAI voice dialogue projects.Experience intelligent voice interaction instantly. (Note: OpenAI functionality requires your own API key.)
- 【STABLE WIRELESS & CLEAR AUDIO】Integrated 2.4GHz Wi‑Fi + Bluetooth 5 (LE); dedicated audio decoding module for natural, responsive voice interaction.
- 【USER‑FRIENDLY VISUAL & PLUG‑AND‑PLAY】2” TFT‑SPI color screen shows real‑time chat; modular design, no extra wiring, ready to use after setup.
- 【FULL LEARNING SUPPORT】45 programmable GPIOs, rich interfaces, online web tutorials, free technical support for beginners & developers.
How to check for this class of silent defect
- Configuration: In tests, compare the requested setting with the value actually sent to the backend. Include non-default values so a silently ignored option is distinguishable from the default.
- Typed filters: Test booleans and numeric representations, including IN and NOT IN behavior. Assert the exact expected document IDs, not only that the query completed.
- Multiple calls: Exercise a workflow with more than one model call and verify that detailed token fields accumulate, as well as checking overall input and output totals.
- Fix status: Before relying on a reported correction, check the upstream pull request or issue and identify the release that contains it. The article reviewed here does not confirm current merge or release status for any of the three reports.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
