Welcome to join.dev
Your home base for developers — watch live streams and talks, track hackathons and releases, follow learning paths, connect your favorite apps, and find your people, all in one activity feed.
Streams, hackathons, learning paths, and more — all in one feed.
🚧 join.dev is a work in progress. We're building in public. 🚧 Building in public. to stay tuned as new features land.
Your feed
Kimi K2.6 lands on Hugging Face with MIT-licensed MoE weights
Moonshot AI's Kimi K2.6 model card is live on the Hub, a 1T-parameter MoE model with a DeepSeek V3-style architecture and native text, image, and video input, released under a modified MIT license.
https://huggingface.co/moonshotai/Kimi-K2.6Models · huggingface · open-weights · moe
BitsMoE: spectral bit-allocation for MoE LLM quantization
New paper proposes an SVD-based spectral-energy-guided bit allocation scheme for quantizing Mixture-of-Experts LLMs, claiming a 12.3x faster quantization pass and 1.76x decoding speedup over GPTQ at 2-bit precision on Qwen3-30B-A3B.
https://arxiv.org/abs/2606.00079Papers · quantization · moe · paper
OpenAI reveals Jalapeño, its custom AI inference chip with Broadcom
OpenAI shared plans for Jalapeño, a custom inference chip built with Broadcom, joining Google, Apple, and SpaceX in reducing reliance on Nvidia for AI compute.
https://techcrunch.com/video/why-everyone-from-openai-to-spacex-is-building-their-own-chips-and-turning-up-the-heat-on-nvidia/AI News · ai-news · chips · hardware
Gemini Embedding 2 goes GA with native multimodal retrieval
Google's Gemini Embedding 2 is now generally available, mapping text, images, video, audio, and PDFs into one embedding space with Matryoshka Representation Learning for flexible 3072/1536/768-dim output.
https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-embedding-2-generally-available/Retrieval · embeddings · multimodal · google
LiveCodeBench v6 leaderboard: Qwen3.7 Max tops coding evals
The contamination-free LiveCodeBench v6 leaderboard now covers 53 evaluated models across code generation, self-repair, execution, and test-output prediction, with Qwen3.7 Max leading at a 0.916 score.
https://livecodebench.github.io/leaderboard.htmlBenchmarks · benchmarks · code-eval · leaderboard
GLM-5.2 beats GPT-5.5 on long-horizon coding benchmarks
Z.ai's open-weight GLM-5.2 outperforms GPT-5.5 on several long-horizon coding benchmarks at roughly 1/6th the cost, a good reminder to weigh cost-per-task alongside raw accuracy when designing eval harnesses.
https://venturebeat.com/technology/z-ais-open-weights-glm-5-2-beats-gpt-5-5-on-multiple-long-horizon-coding-benchmarks-for-1-6th-the-costEvals · benchmarks · evals · coding
Anthropic on context engineering for agents
Anthropic's engineering team argues the discipline has moved past prompt wording alone: the real lever is curating the smallest set of high-signal tokens across the whole context window, not just the system prompt.
https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agentsRAG · rag · context-engineering · agents
TeleRAG: lookahead retrieval cuts RAG inference latency
A paper proposing lookahead retrieval to prefetch likely-needed documents before the generation step even asks for them, trimming end-to-end RAG latency without hurting answer quality.
https://arxiv.org/pdf/2502.20969RAG · rag · latency · retrieval
Hands-on guide: fine-tuning your first LLM with PyTorch + HF
A practical walkthrough of fine-tuning a small open model end to end in PyTorch and Hugging Face Transformers, from tokenization through training loop to eval.
https://huggingface.co/blog/dvgodoy/fine-tuning-llm-hugging-facePyTorch · fine-tuning · pytorch
Databricks guide to picking LoRA rank and target modules
Databricks breaks down how rank, alpha, and target-module choice actually trade off against VRAM and overfitting risk when you're doing LoRA fine-tunes.
https://www.databricks.com/blog/efficient-fine-tuning-lora-guide-llmsFine-tuning · fine-tuning · lora
Loading more…