Updated
Updated
Showing low-relevance items too. Hide low-relevance items
Liquid AI@liquidaiOfficialAI score40AILiquid AI has released its d1 decision models, with d1-3B recommended for high-quality text and vision tasks and d1-omni-600M offered as an experimental option for smaller footprints. The weights are available for download, fine-tuning, and local deployment.
Liquid AI@liquidaiOfficialAI score52AILiquid AI released Open d1, two open-weight multimodal models in its d1 decision model family. The d1-3B model supports text and vision, while d1-omni-600M supports text plus image or text plus audio. The source says the models are meant for real-time decision making across data centers, RTX workstations, and Jetson edge devices.
Epoch AI@EpochAIResearchOfficialAI score10AIEpoch AI published an update to its EBR benchmark, describing methodological changes in a linked article. The post itself provides no further specifics on what those changes are or what they affect.
AILiquid AI released two open-weight decision models, d1-3B and d1-omni-600M (experimental), built on its Liquid Foundation Models and available on Hugging Face. d1-3B scores 48.57 on the Decision Index 0.2.1, the highest among decision models under 10B parameters, and answers a question in 16 ms on an NVIDIA Jetson AGX Thor and under 50 ms on a Jetson Orin Nano. The models support text and images (d1-3B) or text with image or audio (d1-omni-600M).
Aravind Srinivas@AravSrinivasXAI score62AIPerplexity is open-sourcing pplx-embed-v2-late, multi-vector embedding models for text and images in one shared space, in 9B and 0.6B sizes. The 9B model can index multimodal data, the 0.6B model can run queries on device, and PDF pages can be searched without OCR. The author reports 92.4% on MADQA and 64% on BrowseComp+, with weights available on Hugging Face.
Unsloth AI@UnslothAIOfficialAI score40AIUnsloth released an open-source method to fine-tune LLMs into decision models that run locally, lifting Qwen3.5 0.8B's aggregate accuracy from 20.7% to 74.3% across three decision benchmarks. The team used a Clef head with LoRA (r=64) for one epoch on just 4GB VRAM, with the approach applicable to models such as Qwen3.8 and Gemma 4. A guide and notebooks are available on the Unsloth documentation site and GitHub.
Perplexity@perplexity_aiOfficialAI score27AIPerplexity has published both pplx-embed-v2 models on Hugging Face, with the collection available at the linked URL. The post provides no further details on model sizes, benchmarks, or capabilities.
Perplexity@perplexity_aiOfficialAI score45AIPerplexity has released pplx-embed-v2-late, two late-interaction embedding models that retrieve text, images, and pages within a shared embedding space for cross-model querying. The company says both models achieve frontier performance and are publicly available on Hugging Face.
AIMicrosoft Research Asia has open-sourced Agent Lightning v1.0, a roughly 3,500-line agentic RL framework that trains the same agent harness used in deployment. In an end-to-end coding agent pipeline, Qwen3.5-9B rose from 41.8% to 56.4% Pass@1 on SWE-bench Verified using about 6,000 training samples. The framework runs agents as standard Kubernetes jobs without paid commercial sandbox services.
Why it matters: The source shows how training with the deployed agent harness avoids rebuilding agents, and reports concrete SWE-bench Verified gains from about 6,000 samples.
Ai2@allen_aiOfficialAI score20AIAi2 introduces Bolmo, a model that operates on bytes rather than tokens, which the team hopes will lead to larger models adapting how they represent information across languages and domains. Because bytes can also represent images and audio, the approach could eventually extend beyond text.
Alexandr Wang@alexandr_wangXAI score12AIAlexandr Wang announced that the Muse iPad app is now available, praising its slick interface and the generous screen space for users to spread out. He credited Ivan Bruel and the Muse team for the release.
Google Gemma@googlegemmaOfficialAI score41AIGoogle's Gemma account promotes research showing how AI can help scientists secure the global food supply, linking to a Gemma 4 and BOTANIC-1 demo on Hugging Face. The post also offers the BOTANIC-1 download and the accompanying paper.
elvis@omarsar0XAI score44AISmallest AI's Pulse ranks first on the Diarization + ASR track of Voice Arena's Diarization Bench with a 24.4% DER, where lower is better. The next system in that track scores 40.7% DER, and Pulse misses 4.4% of reference speech, the lowest in the track.
AIMistral has released Mistral Large 4, nicknamed Le Chonk, a 1 trillion-parameter model it says can be used and customized by anyone. It is in preview, with a final version due by the end of the month, and is optimized for coding and cyberdefense as well as manufacturing, finance, and electrical engineering tasks. Mistral claims it is the most capable open-weight model developed outside China and says it was trained from scratch rather than through distillation.
AIStacklok, founded by Kubernetes creators Craig McLuckie and Joe Beda, has released Mecatl, an open source cloud-native harness for coding agents on GitHub. Mecatl keeps the agent loop separate from the client, model provider, state store, and execution environment, and moves tool calling, session management, and memory into manageable systems. The article also covers ToolHive, an MCP platform, and an AI Gateway that is not yet open sourced, with a commercial enterprise control plane tying the pieces together.
Mistral AI@MistralAIOfficialAI score16AIMistral Large 4 is the top open-source model on Harvey's Legal Agent Benchmark (LAB), according to Mistral AI's post. The post is a short announcement with no further scores or details.
Teknium 🪽@TekniumXAI score36AIDevelopers are running Hermes on devices such as LilyGO watches, AIPI Lite, desk gadgets, and old Android phones after the Hermes Gadget open SDK and demo were released three days ago. The post credits @NousResearch and says more boards are landing on main through contributor PRs, with the SDK available on GitHub.
Orange AI@oran_geXAI score35AIOrange AI released "podcast-to-article," a free MIT-licensed open-source skill that reorganizes podcast transcripts by topic rather than speaker timeline. The author says it keeps every number and original quote while cutting off-topic tangents, and that early readership and engagement were good.
Soumya Swaraj@isoumyaswarajXAI score13AIA team announces Withicons, a free library of 10,000 icons that all animate, with no sign-up required. The icons come in a single family that AI tools can also use, and the launch is live on Product Hunt today.
🚨 AI News | TestingCatalog@testingcatalogXAI score47AIMistral released Mistral Large 4 "Le Chonk", a 1T-parameter (49B active) multimodal model, with open weights planned in about three weeks. Google rolled out Nano Banana 2.1 across Gemini, AI Studio, and the Gemini API, and released EmbeddingGemma 2, a 740M-parameter open multimodal embedding model under Apache 2.0. OpenAI launched the Decisions API in beta with gpt-6-luna, returning typed answers 10x faster than the Responses API.
AIMeta has open-sourced Rebalancer, a C++ library with a Python interface for solving assignment problems under constraints and objectives, released under Apache 2.0. The article reports that Meta has used it for resource allocation for over 9 years and runs about 40 million problems a day, with P99 solve time of 12 seconds on 265k objects and 3.2k bins. The package can be installed with pip install rebalancer, though PyPI still classifies it as Alpha.
vLLM@vllm_projectOfficialAI score34AIvLLM Semantic Router and KR Labs released Vela 2.0, which makes span-level decisions covering safety checks, domain classification, PII spans, and unsupported claims in a single call. It comes in four sizes from 0.3B to 9B parameters under the Apache-2.0 license.
SemiAnalysis@SemiAnalysis_XAI score7AISemiAnalysis says its open-source ML systems performance dashboard, InferenceX, now includes a GTA theme alongside existing CSGO and Minecraft themes. The post asks followers which theme to add next.
meng shao@shao__mengXAI score62AIGoogle DeepMind released EmbeddingGemma 2, an open 740M-parameter embedding model that maps text, code, images, video, and audio into one 768-dimensional space. Text-only use needs a 270M-parameter footprint, about 191MB active RAM when quantized on a Pixel 11 Pro, while loading all modalities takes about 567MB. The reported MTEB Code NDCG@10 score is 78.68, about 14% above the first generation, and MTEB Multilingual v2 is 61.36, roughly flat.
Nathan Lambert@natolambertXAI score40AIOpenAI is releasing a broad range of new mathematical results produced by an internal frontier model, with the repository hosted at The release was prepared with advice from the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study. The main post itself only comments on the humor of the repository's name.
Simon Willison@simonwXAI score29AISimon Willison released an LLM plugin for sending prompts to OpenAI's new Jev-clone decision model, which is based on GPT-6 Luna. The post links to his write-up with details about the plugin.
AILiquid AI released two open-weight d1 decision models, d1-3B and d1-omni-600M, on Hugging Face. d1-3B scores 48.57 on the Decision Index v0.2.1 public split and answers a single question in 8 ms on an NVIDIA GeForce RTX 4090 and 50 ms on a Jetson Orin Nano. d1-omni-600M is an experimental checkpoint that handles text with images or audio and scores 15.95 on the same index.
Why it matters: The release pairs open-weight decision models with measured latency across Apple, NVIDIA, and Jetson hardware, showing how edge deployment changes what is practical.
AISimon Willison released llm-openai-decisions 0.1a0, a plugin that adds OpenAI's new Decisions API to the LLM command-line tool. The plugin supports yes/no, choices, and score question types, and works with the gpt-6-luna decision model, which accepts both text and image input. OpenAI charges 10 cents per million input tokens for gpt-6-luna, while Jev's rate is 4.2 cents per million, and output is not charged.
Andrew Curran@AndrewCurran_XAI score17AIOpenAI has published a new Math post with a linked repository at The post itself gives no further details in the provided text, so the scope of the work cannot be confirmed from this source.
Nous Research@NousResearchOfficialAI score30AINous Research's Hermes Bench runs every model through the same harness, with reasoning set high where offered. The index reports each model's mean score and mean cost per task across four suites.
AIPenguin Mail 1.0.5 is a free, GPL-3.0-or-later email and calendar app for x86_64 Linux that supports Gmail, Microsoft, IMAP and POP3 accounts. The app includes an optional AI assistant that stays off until a model is chosen and can run locally through LM Studio or Ollama, asking before it sends mail or changes settings.
ollama@ollamaOfficialAI score18AIOllama posted a link to a model page for embeddinggemma-2 in its library, tagging GoogleDeepMind. The post itself gives no further details on the model's parameters, benchmarks, or capabilities.
Fireworks@FireworksAI_HQOfficialAI score28AIFireworks' Head of AI Developer Education fine-tuned a 9B Jev-style classifier on Fireworks using plain SFT, public datasets, and two simple tricks. The full recipe is open source so developers can train their own decision classifiers without a frontier lab budget.
Gamma@GammaAppOfficialAI score42AIGamma announces Gamma 5, a rebuilt version of its presentation platform that it says overhauls how the product thinks, designs, and edits. The company says the release addresses concerns that AI-generated output looked too similar across tools, and it revamps the agent, design tools, editing, import, export, and connectors. Gamma says teams can build presentations, docs, social assets, and graphics that follow their brand or a new aesthetic, using every frontier and image model under the hood.
AIGemini CLI v0.63.0 adds a retry progress indicator during connection recovery and fixes an infinite authentication loop caused by file contention, headless keyring issues, and supervisor state drops. The release also bounds tool output size and cleans up temporary directories when background shell execution exits, alongside fixes for MCP enablement config handling and stdin restoration after capability detection.
AIGoogle DeepMind has released EmbeddingGemma 2, an open 740 million parameter model that maps text, images, audio, and video into one embedding space. It is built on the Gemma 4 architecture under an Apache 2.0 license and supports an 8K token context window. The company reports a code benchmark gain from 68.76 to 78.68 on MTEB Code and says the model can run on-device with about 567MB of active RAM for the full multimodal version on a Google Pixel 11 Pro.
Why it matters: The release shows how a 740M-parameter embedding model can cover text, code, images, audio, and video on local hardware, with memory and storage figures to compare against other on-device options.
dex@dexhorthyXAI score8AIDex Horthy announced that the /show-me skill has reached up to 25k installs. The post gives the install command npx skills add humanlayer/skills --skill show-me.
Modal@modalOfficialAI score36AIScale AI has released AgentEnv, an open source framework for building high-fidelity reinforcement learning environments. Modal says its Sandboxes can deploy these environments in seconds at the scale RL rollouts require.
AIClaude Code v2.1.292 adds a --marketplace option to claude plugin install, which adds the marketplace if needed and then installs the plugin from it. The release also adds an effort parameter to the Agent tool and fixes several security issues, including permission prompts bypassed for network (UNC) file reads and a sandboxed read path that could return files outside approved access.
Philipp Schmid@_philschmidXAI score22AIA Hugging Face Space lets users try EmbeddingGemma 2 directly in their browser using WebGPU. The post shares only a demo link, with no further details on model size, benchmarks, or pricing.