中/EN

Home

Open Source

Curated open-source AI projects with verified GitHub data and hands-on breakdowns.

Frontline HotspotHardcore ReviewsOpen SourceField SOP

Alibaba Open-Sources Qwen3.8-Flash-Next, a Qwen4 Preview

On 2026-08-26 Alibaba open-sourced Qwen3.8-Flash-Next on Hugging Face and ModelScope: a 125B MoE model with 6B activated per token, the first open-weight preview of the Qwen4 architecture. Native context is 262K, extensible to 1M via YaRN; API pricing is \$0.16/\$0.47 per million tokens (about one-twelfth of flagship Qwen3.8-Max). Benchmarks: DeepSWE 58.7, SWE-bench Pro 62.5, CoWorkBench 73.9, AndroidWorld 84.5, MathVision 95.7, and rank 7 on the open Agent Arena. It ships under the qwen-community-1.0 license (not Apache 2.0), permitting commercial use and self-hosting, but verify the terms against the model page before commercial use.

DeepSeek-V4-Flash-Vision-Exp Open-Sourced: MIT Vision Build

DeepSeek-V4-Flash-Vision-Exp is an experimental vision build of V4-Flash. Verification correction: it is not on GitHub (the api.github.com repo returns HTTP 404); the open-source home is Hugging Face (deepseek-ai/DeepSeek-V4-Flash-Vision-Exp). IThome actually reported it on 2026/8/31 19:35, not 9/1. MIT license (commercial use allowed), image input for JPEG/PNG/GIF/WebP, vision "near Claude Opus 4.8" and plain-text parity with V4-Flash — all from official API docs and media. Params ~305B and 1M context are media-reported 【Unconfirmed】. This resource gives landing guidance: self-host via the HF model card, light use via the official API, and smoke-test with real business images first. Confirmed vs unconfirmed items are split in tables.

Harbor and Terminal-Bench: Verify Vendor Scores Yourself

On 2026-09-01 Anthropic published Fable 5.1/Mythos 5.1 Terminal-Bench 4.0 scores — Fable 5.1 55.8%, Mythos 5.1 60.9%, GPT-5.6 Sol 37.3% — and harbor pushed code the same day. This article shows how to use that open-source toolchain to turn vendor scores from "cite-only" into "reproducible". GitHub API measured: harbor-framework/harbor (4872★/1705 fork/Python/Apache-2.0/created 2025-08-04/pushed 2026-09-01) is the eval framework; harbor-framework/terminal-bench (594★/pushed 2026-09-01, most active) is the task and benchmark suite; the old laude-institute/terminal-bench has 301-redirected to terminal-bench-1 (2559★ but stalled at 2026-07-11) — do not conflate them. Terminal-Bench 4.0 recalibrated quotas, removed 8 tasks and fixed 19, so scores are not comparable to earlier versions. A four-step self-built eval: start with private tasks, version the harness config, repeat and report distributions, treat cost and failure modes as first-class.

A 39k-star open-source second brain, but the license is GPL-3.0: OpenHuman in depth

tinyhumansai/openhuman (GitHub API measured 2026-09-01: 39,264 stars / 3,855 forks, Rust, GPL-3.0, created 2026-02-18, 382 open issues, not archived) is a local-first "second brain." Three architectural layers: a scored Markdown-tree memory (SQLite plus an Obsidian mirror you can hand-edit and grep), checkpointed graph orchestration (not loops — pausable and resumable across restarts), and Signal-protocol end-to-end encrypted agent-to-agent comms ("No server ever sees plaintext"). TokenJuice compresses tool output ("up to 80%" is a self-reported ceiling, not a measured average). Privacy Mode is enforced in the Rust core so inference never leaves the machine. Three caveats: Early Beta plus 382 issues means no production SLA yet; GPL-3.0 is strong copyleft, get legal review before embedding in closed source; the README's four-way comparison table is stale (it still labels OpenClaw "Plugin-reliant / Single loop / no Auto-fetch" while 2.0 introduced Shared Cloud Sessions and Active Memory), and the 90k Skills / 5k MCP / 80% compression figures are all vendor-reported, unverified by any third party.

Microduck: Hugging Face's $399 Open-Source Duck Robot Ships a Full RL Stack, With Hardware Files in Dispute

On 2026-08-27 Hugging Face and Pollen Robotics launched Microduck, a 25cm duck-shaped biped weighing under 800g at $399, with preorders opening the same day. The hardware is an RK3566 with 1GB of RAM and 32GB of storage, 15 XL330 servos, an 8x8 ToF LiDAR, and a 2600mAh battery rated at about an hour, running its control policy as ONNX on board at 50Hz, which is the spec that separates it from an app-driven toy. The valuable part is the software released alongside it: runtime and SDK, physics simulation, and a complete reinforcement learning training stack, published across `pollen-robotics/microduck` and `pollen-robotics/microduck_rl`, both Apache-2.0, measured via the GitHub API on 2026-08-31 at 4108 stars / 485 forks and 1037 stars / 179 forks. Training uses MuJoCo Warp (mjlab) with PPO, producing a deployable gait in one to two hours across 4096 parallel environments, with an actuator model, domain randomisation, and backlash simulation bridging sim-to-real, and observation normalisation baked into the exported ONNX. One honest caveat: whether the hardware design files were released too is disputed across sources, so this piece devotes a section to separating the four layers rather than picking a side.

Architecture Innovation Needs Someone to Write the Kernels: Qwen Open-Sources FlashQLA, 2-3x Faster GDN Forward, Hopper and Up Only

QwenLM/FlashQLA (GitHub API, checked 2026-08-30: 670 stars / 69 forks, Python, MIT, created 2026-04-24, last push 2026-08-26) is a high-performance linear attention kernel library built on TileLang. The authors report 2-3x forward and 2x backward speedups for GDN Chunked Prefill over the FLA Triton kernel on NVIDIA Hopper and Blackwell, with the largest gains in pretraining and edge-side agentic inference. Three things worth studying: gate-driven automatic intra-card context parallelism (exploiting the GDN gate's exponential decay under TP, long sequences and small head counts), hardware-friendly algebraic reformulation (cutting Tensor Core, CUDA Core and SFU overhead without losing precision), and TileLang fused warp-specialized kernels (hand-written warpgroup specialization that balances context-parallel and backward needs). Since v0.1.2 it also serves as a GDN backend for flash-linear-attention, plug-and-play through the standard FLA API. The timing is the interesting part: it shipped the same day as Qwen3.8-Flash-Next, meaning Alibaba handed in the architecture paper and the kernel implementation together. The entry ticket is not cheap though - SM90 or newer, CUDA 12.8+, PyTorch 2.8+, and the 2-3x figure is self-reported rather than independently reproduced.

DSH Desktop Teardown: the #1 GitHub Weekly Project Is a Desktop Shell That Turns a 200k-Star Harness into a Double-Click Install

anywhere-labs/dsh-desktop (formerly deepseek-harness-desktop; 21,750 stars / 1,062 forks, MIT, TypeScript, API-checked 2026-08-29) is a native Windows/macOS desktop shell for the 200k-star DeepSeek Harness: it wraps the upstream local Web UI, Host service and plugin system into a desktop app with window, tray, terminal and updates out of the box, under the motto "everything is a plugin, and the desktop itself is a plugin". The week of August 23 it topped the GitHub weekly trending list with +12,488 stars. The README clearly states this is an independent community project with no affiliation with DeepSeek, pins specific upstream versions for stability, and ships a full doc set (user guide, privacy policy, plugin development, ecosystem charter), monetized via sponsors including Alibaba Cloud Wuying, UCloud and 88API. Twelve thousand stars in a week is a case study in ecosystem division of labor: upstream focuses on the harness, the community ships the product.

Open-AutoGLM: Zhipu Open-Sourced a Phone Agent That Actually Clicks - "Open Xiaohongshu and Search for Food" and It Does It

zai-org/Open-AutoGLM (26,104 stars / 4,019 forks, Apache-2.0, Python) is Zhipu's open-source phone GUI agent framework: ADB controls a real device while a VLM reads the screen, looping through screenshot, understand, plan, and tap with 10 fixed action primitives - no accessibility APIs, no widget trees. It pairs with the AutoGLM-Phone-9B model (third-party API, local vLLM with 24GB+ VRAM, or Claude Code + GLM Coding Plan), ships two safety gates (sensitive-action confirmation and human takeover), and officially covers 50+ Android and 60+ HarmonyOS apps. The README includes a whole chapter written for AI assistants to self-install. Caveat: last push was 2026-03-06, so stars reflect accumulated history, not current activity.

GPT-Image 2 Prompts Reverse-Engineered: 538 Cases and 21 Industrial Templates Open-Sourced (20.2k Stars)

#5 on GitHub's weekly trending (+5,329 stars): freestylefly/awesome-gpt-image-2 reverse-engineers scattered community GPT-Image 2 cases into structured prompt assets - 538 cases across 13 categories (UI 73, posters 88, photography 78, illustration 58, infographics 52, e-commerce 41), distilled into 21 industrial templates, each with fill-in-blank, JSON, and pitfall guides. The core thesis is Prompt as Code: compress prose prompts into atomic structures of subject/lighting/materials/layout/text-lock, built for agent and automation reuse. It also ships as an npm Agent Skill (gpt-image-2-style-library) installable into Claude Code and Codex with one command. API snapshot 2026-08-26: 20,193 stars / 2,033 forks, MIT, created 4 days after GPT-Image 2 launched, updated daily for four months.

OpenViking: A Context Database for Agents, Open-Sourced by ByteDance at 33,000 Stars

ByteDance's Volcano Engine open-sources OpenViking: a context database that unifies agent memories, resources, and skills into a viking:// virtual file system you browse with ls/tree/find instead of a black-box vector store. Core designs include L0/L1/L2 tiered loading (saves tokens), recursive directory retrieval, observable retrieval traces, and session-committed memory. 33,172 stars on GitHub (API snapshot 2026-08-25), #6 on the weekly trending chart with +3,540 stars. Official benchmarks: Claude Code long-conversation memory jumps 57.21% to 80.32%, input tokens down 34.3%-91.0%. AGPL-3.0 - mind the viral clause for commercial SaaS.