Knowledge catalogue
Qwen3.8-2.4T-A95B is live on Together AI. The Qwen team’s new open-weight MoE packs 2.4T total parameters with 95B active, a 1M context window, and strong coding & agent capabilities. Start building:
Muse Glimmer is live on Fireworks. The new open-weight model from Meta Superintelligence Labs is a 30B dense model built for always-on agents that reason across many sequential tool calls and can reco
We are in an insane run of open-weight drops. Every modality, open source is winning. This is what an open source AI summer ☀️ looks like: 🧠 LLMs & Reasoning → DeepSeek-V4-Flash-0731 (my king 👑): 304B
OWN YOUR INTELLIGENCE Last year, building on open-weight models was primarily a cost rationalization exercise. Slightly worse performance for a much cheaper price. Now, it is increasingly an existenti
Whoa, Meta released a new open-weight LLM yesterday, something that hasn't happened since the good old Llama days. Their Meta Muse Glimmer model is a 30B multimodal reasoning model with a Gemma-like a
Introducing Muse Glimmer, an open-weight 30B-parameter model optimized for local, always-on agent workflows. Muse Glimmer delivers strong performance on key agentic use cases and benchmarks compared w
arXiv:2506.03259v3 Announce Type: replace Abstract: Purpose: To compare five lightweight open-weight large language models (LLMs) with a rule-based algorithm (RBA) and fine-tuned RadBERT for zero-shot
Agents want to collaborate. So we’re putting that instinct to work on improving open-weight LLMs at formal math. Our latest agent collab tackles the @SAIRfoundation challenge of building a cheat sheet
Security at the foundation requires openness at the foundation. As open-weight models become critical infrastructure for the next generation of software, the systems around them must be transparent, i
arXiv:2607.27421v1 Announce Type: new Abstract: Intent classification is a core component of task-oriented dialogue systems, yet practitioners have limited systematic guidance for selecting deployable
Inkling-Small is now live on Together AI. @thinkymachines’ new open-weight multimodal model delivers similar performance to Inkling at one-quarter the size, built for coding, agents, and general multi
The Kimi K3 architecture figure for yesterday's big open-weight model release, along with some observations and thoughts. 1. Yes, it looks relatively complicated, but it's essentially a scaled-up prod
We recently joined NVIDIA, Microsoft, and others in supporting open-weight AI. Today, we’re putting that belief into the product with model choice on Replit, starting with Kimi K3. The future is the r
arXiv:2607.22929v1 Announce Type: new Abstract: A short fine-tuning run can undo the safety guards of an open-weight model---retraining a refusal-trained assistant to aid weapons development or produc
“An open-weight frontier model helped contain the intrusion.” Attackers have frontier AI. Defenders need a frontier AI ecosystem—the best open and closed models, force-multiplied by a global community
Big update: Among open-weight models, Kimi K3 (Max) is #1 in the Agent Arena with +9.75% net-improvement, surpassing GLM-5.2 (Max) at +7.12%, and landed the #1 spot across 5 signals (see below). Kimi
BREAKING: Inkling by @thinkymachines is 9th overall on Agentic Web App Arena by Design Arena with an Elo of 1257 It's an open-weight model in the same performance band as Claude Opus 4.6 by @Anthropic
Thinking Machines Lab: Thinking Machines Lab debuts Inkling, an open-weight MoE model with 975B total and 41B active parameters, trained to be broad rather than optimized for one area — Try on Tinker
Loving this piece: bringing delight to your daily coffee, your daily workflows, and your AI bill. @gumloop is now running open-weight models on Fireworks in production. 80% lower cost vs. closed-lab A
Julie Bort / TechCrunch: Ollama, which helps developers run open-weight AI models locally, raised a 65M Series B led by Theory Venture, following a 15M Series A led by Benchmark — The popular open sou
Before the week ends, let's acknowledge one of the most INSANE week ever for open AI, with 25+ notable open-weight drops across every modality: 🧠 LLMs → NVIDIA Nemotron 3 Ultra: 550B hybrid Mamba-MoE,
And another open-weight release. Nemotron 3 Ultra has an ultra impressive capability:efficiency ratio! Design-wise, it carries forward the Mamba-2-attention hybrid stack and LatentMoE introduced in th
The MiniMax M2 series was one of the most widely used open-weight LLM series earlier this year. Now, we got a technical report with some interesting tidbits. I summarized some of them below: 1. Full a
arXiv:2605.11255v1 Announce Type: new Abstract: We present Hebatron, a Hebrew-specialized open-weight large language model built on the NVIDIA Nemotron-3 sparse Mixture-of-Experts architecture. Traini
open-weight LLMs have come a long way on agent tasks! but the harness you wrap them in matters just as much as the model itself, and arguably the interface you use to drive that harness matters even m
arXiv:2605.00334v1 Announce Type: cross Abstract: Production agentic systems make many model calls per user request, and most of those calls are short, structured, and routine. This raises a practical
ICYMI: Kimi K2.6 is now in Notion. It’s the first open-weight model that plays like other top-tier models. It’s strong at tool use and carrying out tasks no matter how meticulous or ambiguous. Give it
Kimi K2.6, a language model developed by Moonshot AI, has achieved the top ranking on the open-weight category of the Vals Index benchmark, while placing 7th overall across all model categories. The V
We are entering an extremely exciting era for open-weight models. Kimi K2.6 now feels like a top agentic model. I took it for a spin via @FireworksAI_HQ fast inference APIs. Kimi K2.6 has impressive a
arXiv:2610.09004v1 Announce Type: new Abstract: Does removing harmful information make open-weight models resistant to fine-tuning attacks? We show that mutual information at release alone cannot univ
HuggingFace just closed a major gap in harness engineering! (the ultimate guide to multi-harness RL) the same open-weight model can perform well inside one harness, then lose accuracy or produce inval
arXiv:2609.37902v1 Announce Type: new Abstract: Existing LLM routers choose among models using static per-model costs. We show that open-weight inference markets introduce a second, largely ignored de
Day-0 OpenVINO support from @inteldevs! 🥳 Qwen-Image-2.1 is ready to run optimized on Intel hardware. One open-weight checkpoint for both generation and editing. 👇 We're excited to offer Day0 OpenVINO
arXiv:2609.23999v1 Announce Type: new Abstract: How large language models (LLMs) integrate patient risk with clinical cost tradeoffs remains poorly understood. We investigated how four open-weight LLM
Since there’s no comparison chart on the model page, I asked Perplexity to compare it against some relatively small open-weight models in a similar size range. Here are the results. Upd. Terminal-Benc
arXiv:2609.16519v1 Announce Type: new Abstract: Scientific research increasingly relies on large, heterogeneous data sources, motivating interest in retrieval-augmented generation (RAG) systems that p
Allie Garfinkle / Fortune: Arcee AI, which develops open-weight models in the US, raised a Series B at a 1B pre-money valuation; a source says Arcee raised at least 150M — In 2025, Mark McQuade took a
arXiv:2609.16984v1 Announce Type: new Abstract: Open-weight language models publish the strings their chat templates use to mark turns, roles and tool results, which the tokenizer maps back to the res
arXiv:2609.05241v1 Announce Type: new Abstract: A rapidly expanding ecosystem of actors is removing built-in safety guardrails from open-weight AI models. We profile this ecosystem by identifying key
CommerceAgentBench starts with real commercial demand, and Qwen3.8-Max delivers the strongest overall performance among open-weight models. Let's test Qwen on your real-world workflows! 🔥 Most AI benc
⚡Meet Qwen3.8-Flash, a multimodal MoE and an early preview of the Qwen4 architecture, now open-weight! The production version Qwen3.8-Flash will be available soon via QwenCloud API at just 0.16/1M inp
We benchmarked 20+ open-weight models on easy-to-hard document extraction tasks through ExtractBench. The results are all available on @huggingface 🤗 ExtractBench is a schema-guided extraction benchma
SemiAnalysis: A detailed look at Jalapeño, OpenAI's ASIC developed with Broadcom in 16 months, which beat Nvidia, AMD, and Google chips on multiple top open-weight models — OpenAI's self-designed ASIC
arXiv:2608.22176v1 Announce Type: new Abstract: Large Language Models (LLMs) are increasingly applied to clinical prediction tasks such as in-hospital mortality and readmission from electronic health
// There Is No Neutral Harness // Great work discussing some of the issues in harness evaluation. Twelve open-weight models answer the same 3,679 items from ARC, HellaSwag, MMLU, and TruthfulQA under
arXiv:2608.20569v1 Announce Type: new Abstract: Are frontier models able to introspect about their internal states? Recent work suggests that under certain conditions a complex enough model can audit
#1 open-weight model on Harvey's Legal Agent benchmark! 🙌 Strong enough to handle professional tasks. Small enough to run on your local machine. Qwen3.8-27B is becoming part of your everyday workflows
arXiv:2608.15579v1 Announce Type: cross Abstract: Industrial software-engineering teams increasingly need LLM agents that turn bug reports into correct patches, yet benchmark-scale operation adds long
Alibaba's open-weight AI models accumulated 3 billion global downloads in the past 6 months — more than Meta Llama, more than Google Gemma, more than every Chinese domestic competitor combined. The mo
Simon Willison / Simon Willison's Weblog: Qwen 3.8 27B shows a 17GB open-weight general purpose model can have long context, effective tool calling, strong vision ability, and competent code generatio
I have been running some experiments with smaller open-weight LLMs on multiple-choice questions of Swedish medical licensing exams. On a dataset called MedQA-SWE, GPT-4 scored 84% accuracy in 2024 and
everyone is talking about open-weight models. at @huggingface we're invested in open source AI. like, a lot. and more every year. i pulled the public github numbers for our orgs. good excuse to introd
arXiv:2607.21482v1 Announce Type: new Abstract: Large language models (LLMs) and agents are now widely used tools in code development, with data typically sent to third-party cloud-based models. Their
Ashley Capoot / CNBC: In their letter defending open-weight AI, tech companies urge lawmakers to avoid “premature restrictions” that “stifle competition or drive innovation overseas” — Nvidia, Microso
arXiv:2607.20216v1 Announce Type: cross Abstract: Malware analysis demands rapid interpretation of complex detonation reports spanning filesystem, network, and process behaviours. While large language
Today the Department of Energy (DOE) and Arcee AI announced the development of Genesis-Science-1 (GS1), an open model for scientific research. This is a joint effort to bring advanced AI into scientif
Today we’re releasing Poolside Laguna S 2.1 It is a 118B-total, 8B-active open-weight model built for agentic coding and long-horizon work, with context up to 1M tokens https://poolside.ai/blog/introd
With our internal coding benchmark, we're able to confidently introduce open-weight models into our AI code reviewer w/o degrading code quality. Have the frontier model (Fable) to the hardest work, de
'At this very moment China is giving its AI technology away. It's releasing open-weight AI models that are cheap, capable, and they're fast becoming the world's default.' We can overcome this. @neil_c
This post from DAIR.AI recommends resources for developers and organizations scaling applications using open-weight language models, emphasizing the importance of planning infrastructure and deploymen