Marktechpost AI
@Marktechpost • 11,789 subscribers
The fastest AI dev news engine on X — model releases, tools, and what they actually mean
Shorts
![[Most robots react. This one thinks a step ahead.] Ant Group's Robbyant just published LingBot-VA 2.0 — a video-action foundation model built from scratch for robot control, not fine-tuned from a video generator. The usual approach takes a video generator made for content creation and bolts a robot policy onto it. LingBot-VA 2.0 argues that's the wrong starting point, and pretrains the whole causal stack natively instead. What stands out: → Foresight Reasoning — the robot predicts the next action chunk while executing the current one, then overwrites the imagined frame with the real observation. Prediction and execution stop waiting on each other. → 927 ms → 142 ms per chunk, across four cumulative optimizations. That lifts asynchronous control from 35 Hz to 225 Hz — a 6.5× speedup. → One shared latent space. A semantic visual-action tokenizer puts world states and actions in the same coordinates, so unlabeled web video carries action-relevant signal. → Sparse MoE video stream — 128 experts, top-8 routing. Roughly 2.5B of ~15.3B parameters fire per token. → Few-shot by design — adapts from 10–15 demonstrations, and a human demo video can replace the text instruction entirely. Full breakdown: Paper: Project Page: Robbyant Ant Group](https://image.24vids.com/tw-2075856416116772924/amplify_video_thumb/2075855720936005632/img/osBfB6B00BcSMDww.jpg)
Videos
What does "trust remote code" actually approve? In most local AI tools, the answer is: that repo, indefinitely. That matters more now. In May, HiddenLayer reported a trending Hugging Face repo whose loader. py downloaded an infostealer on Windows. JFrog counted 495 malicious models that month. I spent some time with Unsloth Studio's approach. It handles it differently. Approval is tied to a fingerprint of the scanned code. Change the code, and you're asked again. Critical findings can't be approved. Code it can't scan is blocked. There's no trusted-org bypass. Weights are checked on their own. Studio reads Hugging Face's malware verdict without unpickling the file. Flagged files in the loading path are blocked, including weight shards. .bin weights load with weights_only=True on PyTorch 2.6+. The supply chain gets attention too. PyPI and npm packages are content-scanned. npm packages need to be at least 7 days old. GitHub Actions are pinned, and llama.cpp binaries are checksum-verified. Trivy isn't installed at all, after a compromised Trivy exposed LiteLLM's publishing credentials in March. Accounts are covered as well. Passwords use PBKDF2-HMAC-SHA256, failed logins are rate-limited, and the first admin password is random. Managed accounts can't run repository code. For remote access, --secure serves HTTPS through a Cloudflared tunnel. --disable-tools is recommended when exposing Studio. What it doesn't do: sandbox approved model code. Server-side tools still run as your OS user. Local folders aren't covered by the file check, and a pending scan status doesn't block. That's a reasonable line to draw, as long as you know where it is. Read our full analysis on this: Here is the Unsloth's docs: Unsloth AI Unsloth
Marktechpost AI534,474 Aufrufe • vor 11 Tagen
Datalab Introduces OmniExtractBench to Fix Bias and Opacity in Extraction Benchmarks 1 missed table row can drop an extraction score from 99% to 0%. Datalab just open-sourced OmniExtractBench to fix that. 620 documents, 4 vendor benchmarks, 1 deterministic scorer. Here's how it works. 👇 1. Pool the benchmarks 329 docs from LlamaIndex's ExtractBench, 47 from micro1, 42 from Extend's LongArray, 202 synthetic from Datalab. → Forms, filings, decks, 100+ page tables 2. Pair rows by content, not position Hungarian matching aligns table rows before anything is scored. → 100-row table, first row missed: 0% by position, 99% by content 3. 6 verdicts per value matched, misread, unfound, fabricated, invented_item, invented_field. → Every lost point traces back to 1 value 4. Nulls earn nothing Empty strings and None are dropped before scoring. → 100 padded empty fields add 0 verdicts 5. The numbers (10 configurations) → Datalab accurate: 93.85 → Datalab balanced 93.48 vs Reducto 93.47, effectively tied → GPT 5.6-sol: 95.11 precision, 84.99 recall Full analysis: GitHub Repo: Technical details: Datalab Vik Paruchuri
Marktechpost AI392,888 Aufrufe • vor 17 Tagen
This World Model 'LingBot-World-Infinity (LingBot-World 2.0)' just released from Ant Group looks realy promising. It is an open causal world model with an Agentic harness. Most interactive world models hold together for a few minutes. Then textures smear and geometry warps. That's a video model, not a world — and Robbyant just drew the line at the attention mask. They released LingBot-World-Infinity (LingBot-World 2.0) — a 14B open causal video world model built on Wan2.2, trained with a Mixture of Bidirectional and Autoregressive (MoBA) attention mask, then distilled into a few-step real-time generator with no post-hoc drift filtering anywhere in the stack. Here's what's actually interesting: → Pure teacher forcing overfits — as context grows, the model leans on context instead of predicting frames. MoBA appends a bidirectional full-attention block as a regularizer → Leak-free cross-attention: AR rows attend to background prompt a_B plus chunk prompts a_≤i, lower-triangular. Bidirectional rows see one global prompt a_G → DMD runs over long self-rollout trajectories, not teacher-forced states — the student is optimized on the distribution its own errors induce → Director-Pilot harness: a VLM proposes event cards, the DiT generator renders physical dynamics. Mode B adds a SAM tracking loop for object-centric interaction → One 60-minute uninterrupted session, 20 distinct scenarios, no perceptible decay Full analysis: Paper: Model weight: GitHub Repo: Project: Ant Group Robbyant
Marktechpost AI233,449 Aufrufe • vor 3 Monaten
Meet WebBrain: An Open-Source, Local-First AI Browser Agent That Reads Pages and Automates Tasks in Chrome and Firefox WebBrain lives inside your browser and can run entirely on your own local model — no cloud, no account, no data leaving your machine. Most "AI browser agents" are a chat box that pastes your page into someone else's server. That's not an agent that lives where you browse — and WebBrain draws a very clear line between the two. It's an open-source (MIT), local-first browser agent for Chrome and Firefox. It runs inside your existing authenticated session, on a model you pick — so with llama.cpp or Ollama, nothing leaves your machine. Here's what's actually interesting: → Two modes, cleanly separated. Ask reads the page (read-only, content scripts). Act clicks and types through the Chrome DevTools Protocol (chrome.debugger) — trusted input events that modern sites honor, reaching cross-origin iframes and shadow DOM. → UI-first by design. For anything that submits, sends, or buys, it drives the visible UI and refuses to hit REST/GraphQL endpoints directly. It starts read-only and asks before consequential actions. → Bring any model. llama.cpp, Ollama, LM Studio, vLLM — or OpenAI, Claude, Gemini, DeepSeek, Groq, OpenRouter. Recommended local: Qwen 3.6 35B (Qwen3.6-35B-A3B), which beat Gemma 4 on the project's screenshot benchmark. → Tuned for cost and privacy. Token-conscious screenshots, oldest-first context trimming, a dedicated vision model, 40+ tools (~20 in Compact mode). No telemetry. No accounts. Full analysis: GitHub Repo: Chrome Extension: Firefox Add-on: Portal:
Marktechpost AI203,011 Aufrufe • vor 3 Monaten
Soofi Consortium Releases Soofi S 30B-A3B: An Open 31.6B Model for German and English Hitting 79.1 German Aggregate With Only 3.2B Active Parameters. Here's how it works. 👇 1. Sparsity in two places at once 52 layers: 23 Mamba-2, 23 granular MoE, 6 Grouped-Query Attention. The MoE router picks 6 of 128 experts per token, plus 2 shared. Mamba-2 carries the sequence mixing with a fixed-size recurrent state, so 46 of 52 layers keep no KV cache at all. → 3.2B of 31.6B parameters active per token 2. Reference architecture on purpose No bespoke backbone. It adopts NVIDIA's Nemotron 3 Nano design without modification — for day-one vLLM kernels, for serving efficiency, and for scientific control. That last one is the real move: Nemotron becomes an architecture-identical baseline, so the data recipe is the only variable left. 3. German as the deliberate variable Three-phase Warmup–Stable–Decay curriculum. Phase 1 is breadth at a 1e-3 plateau, Phase 2 concentrates high-quality data as the LR decays, Phase 3 stretches context to 1M tokens. → ~26.68T consumed tokens → German 7.2% → 15.32% of the mixture, vs ~5% for all non-English in the Nemotron reference → +4.2 German aggregate, +1.8 English, +6.7 held-out English over Nemotron 4. Where the architecture pays: memory bandwidth Every decoded token re-reads the weights and, for a Transformer, the attention cache of every sequence in the batch. Six KV layers instead of 52 keeps that per-sequence state small. Measured on one B200, TP=1, vLLM latency-subtraction. → 8–9× aggregate decode TPS/GPU vs dense 14–24B models at 40K context, batch 32 → decode stays flat from 4K to 256K 5. The numbers (base model, lm-evaluation-harness, 16 open baselines) → 70.1 English aggregate, +2.8 over Olmo 3 32B → 79.1 German aggregate, +6.3 over Apertus 70B → 73.8 HumanEval, 84.2 MBPP-DE, 88.8 GLP-DE, 61.2 INCLUDE-DE Full analysis: Paper: Technical details:
Marktechpost AI65,637 Aufrufe • vor 3 Monaten
Mistral AI Releases Leanstral 1.5: An Apache-2.0 Lean 4 Code Agent Model Solving 587 of 672 PutnamBench Problems Most AI theorem proving is a language model generating a proof in one shot, with a verifier bolted on at the end to check it. That's autocomplete with a grader — and Mistral just drew a clear line between that and an actual proof agent. They released Leanstral 1.5 — a 119B MoE with 6.5B active parameters, trained as a code agent that lives inside the Lean 4 compiler loop: propose a proof, read the compiler's goals and errors, refine, repeat until it compiles or the budget runs out. Verification isn't the eval here. It's the training signal. Here's what's actually interesting: → Test-time scaling behaves like a dial: PutnamBench Pass@8 climbs 44 → 244 → 493 → 587 solved as the per-attempt token budget moves 50k → 200k → 1M → 4M → 587/672 on PutnamBench at ~$4 per problem, versus an estimated $300+ for Seed-Prover 1.5 high (a 10 H20-days-per-problem budget) → Saturates miniF2F: 100% on both validation and test sets → Two RL environments in training — a multiturn prover, and a raw-filesystem code agent that edits files, runs bash, and queries the Lean language server for live goals and types → Not just math: an Aeneas (Rust → Lean) pipeline flagged 11 genuine bugs across 57 repos, 5 previously unreported — including an integer overflow in datrs/varinteger when (value + 1) hits Std.U64.MAX Apache 2.0 weights, free API endpoint Full analysis: Model weights: Project: Technical Details: Mistral AI Mistral AI for Developers Sophia Yang, Ph.D.
Marktechpost AI56,695 Aufrufe • vor 3 Monaten
Mistral AI Releases Robostral Navigate: An 8B Model Enabling Robots to Navigate Complex Environments Hitting 76.6% on R2R-CE With One RGB Camera. No LiDAR. No depth sensor. No multi-camera rig. Here's how it works. 👇 1. Pointing, not metric commands The model predicts the pixel coordinates of the next target in the camera view, plus the arrival orientation. Working in pixel space keeps it robust to camera intrinsics and world scale. When the target leaves the frame, it falls back to local displacements ("2m forward, 1.5m left, turn 25°"). 2. Grounding-first No open-source VLM base. It starts from Mistral's grounding model (pointing, counting, localization). Navigation emerges once the model knows where things are. → ~400,000 trajectories across 6,000 simulated scenes 3. Prefix-caching for training A tree-based attention mask packs a full episode into one sequence — all time steps in a single forward pass. → 22× fewer training tokens; months of training done in days 4. Online RL on top After supervised training, CISPO adds trial-and-error learning to fight distribution shift from behavior cloning. → +3.2% success rate from RL alone 5. The numbers (R2R-CE, Matterport3D) → 76.6% success on validation unseen → +9.7 pts over best single-camera approach → +4.5 pts over best depth/multi-camera system The key takeaway: state-of-the-art continuous VLN without a sensor stack — grounding-init, pixel-space actions, prefix-cached SFT, and online RL, on one RGB camera. Full analysis: Technical details: Mistral AI Mistral AI for Developers
Marktechpost AI39,955 Aufrufe • vor 3 Monaten
NVIDIA AI Releases Nemotron 3 Embed: An Open Embedding Collection Whose 8B Checkpoint Ranks #1 on RTEB Most RAG stacks treat the embedding model as a commodity — pick one, index, move on. Nemotron 3 Embed is NVIDIA's argument that the retrieval layer is where agent cost actually gets set. They released three open checkpoints — Nemotron-3-Embed-8B-BF16, 1B-BF16, and 1B-NVFP4 — built on Ministral bases, trained with bidirectional attention masking, pooled by averaging token-level representations, all taking 32,768-token inputs under OpenMDW-1.1. Here's what's actually interesting: → The 8B ranks #1 overall on RTEB: 78.46 avg NDCG@10, alongside 75.45 on MMTEB Retrieval and 60.60 on ViDoRe-V3 text → The 1B wasn't trained small. It was pruned from a 3B parent with ModelOpt mcore_minitron NAS, then distilled from the 8B teacher on COS + MSE loss — twice → That pipeline lands the 1.14B checkpoint at 72.38 RTEB, up 10.4 points on the prior-generation llama-nemotron-embed-vl-1b-v2 → NVFP4 costs 0.38 RTEB points (72.00 vs 72.38, ~99.5% retention) and buys up to 2x BF16 throughput on Blackwell Full Analysis: Model weight: Technical details: NVIDIA AI NVIDIA
Marktechpost AI32,352 Aufrufe • vor 3 Monaten
PrismML Releases Bonsai 27B: 1-bit and Ternary Builds of Qwen3.6-27B Hitting 89.5% of FP16 at 3.9GB. No new pretrain. No higher-precision escape hatches. No multi-GPU rig. Here's how it works. 👇 1: Codes, not floats Every weight becomes a code, with one shared FP16 scale per group of 128. Ternary is {−1, 0, +1}, binary is {−1, +1}. Sharing the scale across 128 weights keeps its cost at 16/128 = 0.125 bits. → Ternary: log2(3) + 16/128 ≈ 1.71 bits/weight → 5.9GB → Binary: 1 + 16/128 = 1.125 bits/weight → 3.9GB 2: Post-training, not from scratch No BitNet-style low-bit pretrain. It starts from off-the-shelf Qwen3.6-27B, architecture unchanged. The representation runs end to end across embeddings, attention projections, MLP projections, and the LM head. → 9.4× (ternary) and 14.2× (binary) vs the 54GB FP16 baseline 3: Labels are not bit-widths Conventional low-bit builds are mixed-precision by construction. The advertised name describes the most-compressed tensors, not the model. → Q4_K_XL, labeled "4-bit," is really 5.2 bits/weight at 17.6GB → IQ2_XXS, labeled "2-bit," is really 2.8 bits/weight at 9.4GB 4: Fitting a phone is two budgets iOS caps a single app near half of RAM, so a 12GB iPhone exposes ~6GB. The KV cache grows on top. Hybrid attention at ~75% linear means only 16 of 64 layers cache. → 4-bit KV: 4.3GB at 262K context, down from 17.2GB → 11.0 tok/s on iPhone 17 Pro Max 5: The numbers (15 benchmarks, thinking mode) → Ternary: 80.49 avg at 5.9GB — 94.6% of FP16 → 1-bit: 76.11 avg at 3.9GB — 89.5% of FP16 → IQ2_XXS falls to 57.5 on AIME26 while still scoring 88.93 on MMLU-Redux The key takeaway: 27B-class reasoning without the 54GB checkpoint — group-wise ternary and binary codes, an end-to-end low-bit language stack, 4-bit KV, on one phone. Full analysis: Repo: Model weight: Technical details: PrismML
Marktechpost AI31,860 Aufrufe • vor 3 Monaten
NVIDIA Releases Audex (Nemotron-Labs-Audex-30B-A3B): A Unified Audio-Text LLM That Preserves the Text Intelligence of Its Backbone Most unified audio models pay a text tax. Add audio output, and reasoning benchmarks drop — even when the only new output is speech. NVIDIA just released one that doesn't. Audex (Nemotron-Labs-Audex-30B-A3B) is a 30B MoE with 3B active parameters, built on the text-only Nemotron-Cascade-2-30B-A3B backbone. Audio inputs are projected into the text embedding space. Text tokens and quantized audio tokens are then generated the same way, inside one MoE decoder — no thinker–talker split, no stacked cascade. Here's what's actually interesting: → One model, audio in and out: understanding, ASR, translation, TTS, text-to-audio, speech-to-speech → Text holds vs its own backbone: IMO AnswerBench 81.1 vs 79.3, MMLU-Redux 86.4 vs 86.3 → Beats text-only Qwen3.5-35B-A3B on several tasks: LiveCodeBench v6 85.3 vs 74.6, IFBench 77.8 vs 70.2 → The usual tax, for contrast: Qwen3-Omni-30B-A3B-Thinking drops to 60.4 on HMMT vs 71.4 for its text backbone → 6.82 WER on OpenASR, ahead of Step-Audio-R1.1-33B (7.91) and Qwen3-Omni-Thinking (8.00) → Two codecs: X-Codec2 for speech (50 tok/s, FSQ, 65,536 codebook), X-Codec for general audio (200 tok/s, 4 flattened RVQ layers) — the only strong open model generating general audio beyond speech Full analysis: Paper: Model weights: NVIDIA AI NVIDIA Wei Ping
Marktechpost AI28,301 Aufrufe • vor 3 Monaten
Most speculative decoding still drafts tokens one at a time. That's not parallel generation — it just hides the serial loop behind a smaller model. UC San Diego's z-lab just drew a clear line between the two. They released DFlash — a lightweight block diffusion model that drafts a whole block of tokens in a single forward pass, then lets the target model verify the block in parallel. Up to 15× higher throughput for gpt-oss-120b on NVIDIA Blackwell. No token-by-token drafting anywhere in the speculative path. Here's what's actually interesting: → The drafter is conditioned on the target model's own hidden features, injected into the Key/Value cache of every draft layer — so acceptance length scales with draft depth instead of diluting away → A 5-layer drafter replaces the 7B diffusion drafters that capped earlier methods near 3–4× → MATH-500 speedup: 6.08× vs. 1.81× for EAGLE-3 (4.86× average vs. 1.76×, Qwen3-8B, greedy) → Up to 15× higher throughput for gpt-oss-120b on NVIDIA Blackwell — at the same interactivity target → Lossless: the target still verifies every token, so output quality is preserved Full analysis: Paper: NVIDIA's metrics: Project: Model weights: Repo: NVIDIA AI NVIDIA AIDev
Marktechpost AI23,328 Aufrufe • vor 3 Monaten
Unsloth AI and NVIDIA are Revolutionizing Local LLM Fine-Tuning: From RTX Desktops to DGX Spark Fine-tune popular AI models faster with Unsloth on NVIDIA RTX AI PCs such as GeForce RTX desktops and laptops to RTX PRO workstations and the new DGX Spark to build personalized assistants for coding, creative work, and complex agentic workflows. The landscape of modern AI is shifting. We are moving away from a total reliance on massive, generalized cloud models and entering the era of local, agentic AI. Whether it is tuning a chatbot to handle hyper-specific product support or building a personal assistant that manages intricate schedules, the potential for generative AI on local hardware is boundless. However, developers face a persistent bottleneck: How do you get a Small Language Model (SLM) to punch above its weight class and respond with high accuracy for specialized tasks? The answer is Fine-Tuning, and the tool of choice is Unsloth. Unsloth provides an easy and high-speed method to customize models. Optimized for efficient, low-memory training on NVIDIA GPUs, Unsloth scales effortlessly from GeForce RTX desktops and laptop all the way to the DGX Spark, the world’s smallest AI supercomputer...... Full analysis: NVIDIA NVIDIA AI NVIDIA AIDev NVIDIAnewsroom Unsloth AI Unsloth
Marktechpost AI Dev News ⚡31,551 Aufrufe • vor 10 Monaten
MOONMATH 🔥 : Open-sourced a bf16 forward attention kernel for AMD MI300X, written in HIP instead of hand-tuned assembly. Beats AMD's own AITER v3 on every shape and every rounding mode — geomean 1.18×/1.15×/1.08×, up to 1.26× across an 8K–128K sweep. Core trick: one-instruction asm wrappers pick the exact opcode while the compiler still handles register allocation. Already wired into SGLang diffusion — 1.23× faster Wan2.1 video generation on MI300X, no visible quality regression. MIT-licensed 👀 Full analysis: Technical details:
Marktechpost AI11,165 Aufrufe • vor 3 Monaten
Keine weiteren Inhalte verfügbar