A 35B open-weight model trained to search a precomputed... index answers repo search questions at 100x lower cost than a frontier model. We partnered with turbopuffer to train Qwen3.6-35B-A3B to find code across ~9,000 repositories. It tops the needle-in-a-haystack task outright at 2-10x lower latency.show more

We partnered with Prime Intellect to build Fast Ask,... a small RL-trained subagent that helps our Sheets agent find answers in spreadsheets. It scores +4% over Opus on exact match accuracy at Haiku latency.show more

🥊 Qwen3.6 35B A3B vs Qwen3.6 27B Made them... fight on the same prompt 🌊 Ocean Waves, a canvas challenge 35B took 39.7s (at 142tok/s) 27B took 111.6s (at 50tok/s) Both models are really good. The fact that the 35B produced such a strong result in under 40 seconds is seriously impressive. But when you look at 27B's output... it's actually so much better Added clouds, beautiful foam particles and splashing effects, a perfectly straight horizon, more realistic blinking stars, and a nice sun reflection on the water 🏆 Qwen3.6 27B slower but the result slaps Absolutely amazing that both models run smoothly on 16GB VRAM. Used Unsloth AI GGUFs here, Q3_K_S quant to make it fit in the hardware 🔥show more

Code LLaMA is now on Perplexity’s LLaMa Chat! Try... asking it to write a function for you, or explain a code snippet: 🔗 This is the fastest way to try AI at Meta’s latest code-specialized LLM. With our model deployment expertise, we are able to provide you with this model less than 24 hours of it’s release. What’s next? We’ll integrate code LLaMA into Perplexity, all in service of providing you with the best answers to your most technical questions!show more

HydraFusion Explained. Part I: How does the Copilot engine... know what to optimize for? Your prompt is evaluated across 4 dimensions: ➡ Does it require deep reasoning? (aka. reasoning depth) ➡ Is it a sophisticated problem? (aka. code generation complexity) ➡ Is it untangling a complicated mess? (aka. debugging difficulty) ➡ Is it dominated by tool-use? (aka. tool orchestration needs) Based on this evaluation, a HyDRA score is assigned to determine the capability profile your task needs the most and to establish a quality bar. Part II: How does it choose a model? Note: It doesn' t pick one model to handle the entire job e2e, (that's Auto mode). Instead, it selects 1 of 3 execution workflows and assigns the best model at different stages based on the HyDRA score: 1️⃣ Single ⚙️ How it works: A single model completes the task from start to finish. ⚖️ Rationale: The task comfortably meets the quality bar with one model. Multi-model orchestration would add latency and cost with no meaningful quality gain. 2️⃣ Cascade ⚙️ How it works: A lightweight, cost-efficient model generates the solution. This draft is evaluated against a quality gate and if it falls short of the quality bar, the entire task escalates to a stronger, frontier model. ⚖️ Rationale: Only bring in the big guns when there is concrete evidence that a lightweight model won't meet the quality threshold. 3️⃣ Critique ⚙️ How it works: A lightweight model drafts the initial code and tool interactions. An independent, read-only frontier model reviews that draft and provides feedback. The original lightweight model then performs any targeted revision(s) before the final response is sent to the user. ⚖️ Rationale: Writing code (output tokens) is expensive while reviewing code (input tokens) is cheap. Instead of incurring the cost of a powerhouse writing hundreds of lines from scratch, a cost-efficient model writes the first draft, and the frontier model just reviews it and points out fixes. HydraFusion is available in experimental preview on the GitHub Copilot CLI: /experimental on, /model and select Hydrafusion (Research Preview)show more

A search engine has many moving parts, and they... constantly change with scale of index and traffic. In this blog post, we dive into the code-level abstractions we've built to develop search infrastructure that is both performant and a joy to work with.show more

1/ Happy to share UniDisc - Unified Multimodal Discrete... Diffusion – We train a 1.5 billion parameter transformer model from scratch on 250 million image/caption pairs using a **discrete diffusion objective**. Our model has all the benefits of diffusion models but now in multimodal space! - flexible compute-quality tradeoff, zero-shot inpainting and editing, better control via classifier-free guidance and lower latency! We open source everything - our code, weights and the training dataset.show more

Routing for long-horizon coding agents is a big deal.... Not Diamond just announced a model router that works natively with Claude Code. This is huge. It picks the model and reasoning effort before each turn in a session, runs through a privacy-preserving local proxy, and your requests still execute through your own gateway. In their benchmarks, it approximates Opus 4.8 Xhigh quality at 39 to 61 percent lower cost.show more

Monitor and control your AI spend on every provider... on Our early users save 40% on average. Every week, the price-intelligence-latency frontier shifts, and we expect this trend to continue. Tradeoffs between latency, reasoning, cost, service tier, open source and closed source models are shifting constantly. Router sends every request to the model that's actually best for the task and helps you control what tokens you buy. We benchmark it against real work: ~40% lower cost for the same outputs. Today we're opening it to everyone. Two lines of code or just change your base URL. No Ramp account needed. Free through 2026, first $26 on us. Get an API key today atshow more

5 days ago it took 2 GPUs to build... this. today it takes 1. same prompt. same particle simulation. completely different model. Qwen-Coder-Next (80B) on 2x 3090s. 46 tok/s. 564 lines. 2 iterations to get it working. 48GB VRAM across two cards just to hold it. Qwen3.5-35B-A3B on a single 3090. 112 tok/s. 461 lines. first try. cleaner code, fewer lines, better structured. 19.7GB on disk with 4GB VRAM to spare. half the parameters. one GPU instead of two. 2.4x faster. and the output actually improved. this is what happens when architecture catches up to ambition. Gated Delta Networks(Mamba2 variant) hybrid with sparse MoE. 3B active params out of 35B per token. efficiency at the architecture level, not just quantization. the curve isn't flattening. it's steepening.show more

We took a 30B model and split it in... two to write tokens in parallel instead of one at a time. Introducing Nemotron-Labs-TwoTower: a diffusion language model from NVIDIA Research adapted from Nemotron-3-Nano-30B-A3B. Here’s how it works: one half holds the context, the other writes the tokens, with both reusing the pretrained model instead of training a new one from scratch. We found it kept 98.7% of the original model’s quality at 2.42× faster generation.show more

Robotics keeps hitting the same wall. Single task RL... works, but... it does not scale to hundreds of tasks or new embodiments. This new paper looks like a real step toward fixing that. The team introduces MMBench, a benchmark with 200 tasks across many domains and robots, and Newt, a language conditioned world model trained online across all 200 tasks at once. The simple idea behind Newt: The model learns from demos to get the right priors It trains across many tasks through online interaction It uses language to ground the goal It adapts fast when a new task shows up What stood out to me: ✅ One model trained on 200 tasks at the same time ✅ Language conditioned control for both states and RGB ✅ Better data efficiency than strong baselines ✅ Strong open loop control ✅ Fast adaptation to new tasks and embodiments ✅ Full release of 200 checkpoints, 4000 demos, code, and benchmark This is a good push toward general control instead of one model per task. If you want the full paper: Project page: —- Weekly robotics and AI insights. Subscribe free:show more

let me save you 3 hours of head scratching.... if you're running local models like Qwen3.5-35B-A3B through Claude Code via llama.cpp's Anthropic endpoint, the chain will break every 3 to 5 minutes. tool call fails. flow stops. you reprompt. it recovers. 2 minutes later it stops again. the model is fine. the harness chokes on local inference latency. switch to OpenCode. same localhost endpoint. same model. same GPU. the chain doesn't break. the tradeoff: OpenCode sometimes loops. the model forgets what it already read and repeats the same tool call. but a loop you can interrupt. a broken chain kills your momentum and you start over. watch both side by side. proprietary agent vs open source agent. same 3B model. different failure modes. pick your poison.show more

New open-source agent harness just landed! I got early... access to TrueForge by TrueFoundry and have been running it locally for the past few days. The harness layer deserves as much attention as the model, and open source matters here because you can inspect the loop, run it on your own infrastructure, and swap to the latest or cheaper models. TrueForge handles the runtime work that makes an agent reliable. It drives the tool-calling loop, manages context, coordinates subagents, and executes code in a sandbox, with any model you choose. Every tool call re-sends the growing context to the model, so in practice the harness controls most of what an agent costs to run. A few things stood out from my testing and their published benchmarks. Vendor-Neutral by design. It runs OpenAI, Anthropic, and Google models alongside open-weight models like Kimi, GLM, and DeepSeek. Model routing is a setting, and you can send each task to the model that fits it. On a 14-task enterprise agent benchmark, it matched the accuracy of Claude Managed Agents running the same Opus 4.8 model at roughly 30% lower cost per run (3.8M tokens vs 10M for the same answers). Routing the same tasks to GLM-5.2 held accuracy and brought cost down by about 75%, around $3 per run instead of $12. Fully self-hosted and Open Source (MIT License). I had it running locally with one command, with sandboxed code execution working out of the box. It's time to own your agent harness. Thanks to TrueFoundry for partnering on this post.show more

excited to release RadiusNet-31P, a SOTA open-weight model capable... of predicting the perfect concentric border radius with 99.999% accuracy trained for 5,000 epochs so designers never have to subtract again ↓ open weights in the first replyshow more

Very proud to see the FourCastNet model that I... helped build at NVIDIA in collaboration with Berkeley Lab and other universities be highlighted in Jensen's keynote at #GTC24 - FourCastNet was the very first high-resolution #AI model to show competitive performance for weather forecasting and is part of Earth-2 services. It is now running at ECMWF and you can get 2-week forecast there in realtime. Hear Bill Collins from Berkeley Lab leverage FourCastNet for studying extreme weather events, offering accurate and efficient results from huge ensembles at a lower computational cost.show more

Something just launched that makes VS Code feel ancient.... And im all in for it: Cline Desktop app, an open-source app for open-weight models (and you know how much i love open source) brings its coding agent into a standalone Cline Desktop app for Mac + Windows. Just open a project and tell it what you want to change. No need to open VS Code first. In Cline's demo, it adds priority filters to a task board, works through the code and runs the build. Glad to see the model choice stays open, too. You can bring your own API keys, use supported open-weight models and even switch models halfway through a project!show more

Diffusion for virtual worlds: we've trained a new model... to create 3D objects from text. And it's 50x faster than any alternative. We made each of these cute characters in just one minute with Genmo's new text-to-3D model. Build your own world atshow more

I explored a further possibility with local models: Qwen3.6... 35B A3B + NVIDIA LocateAnything-3B as a local Computer Use agent (proof of concept). In the demo, I asked it to switch my Mac to light mode. It did. Then back to dark. Did that too — finding the right toggle in System Settings, clicking it, and verifying the change itself. It's fully screenshot-based, so no Accessibility API needed. If it's on screen, the agent can see it and act on it. This runs entirely on your own hardware — private, local, built from two small open models.show more

Finding 1: Harness affects cost more than correctness. Fable... 5 can cost twice as much for a 1.1-point gain in success rate. On SWE-bench Lite: - Claude Code: 97.8% accuracy, $1.33/rollout - Pi: 96.7% accuracy, $0.67/rollout While Claude Code reaches the highest success rate on the SWE-bench Lite frontier, Pi and Codex often achieve similar success rates at lower cost across the models we test. So you may be paying a hidden “harness tax” if you pick your harness based only on the success rate… 💸 (2/n)show more

Grok 4.5 is sitting at #2 on the FrontierSWE... leaderboard. Above Claude Opus 4.8. Above GPT-5.5. So yes, the coding model conversation just got a little more crowded at the top. Strong performance is one thing. Doing it with serious speed, better token efficiency, and lower cost is where it gets annoying for everyone else. Builders love a smart model. SpaceXAI Grok X Freeze / Writer: Annette, Designer: Jannéshow more