TechCrunch · AISame story · 2 outlets7171Recursive releases its first model, Flint, a free coding model that can run on laptops, with almost the entire development process driven by AI. Recursive's automated research system, Eureka Machine, participated in almost every development step of Flint, enabling the team to complete the model construction in less than four months.
Hugging Face BlogQwen3-4B's specialized vocabulary list VocabX reduces inference cost by 77% by focusing on the cost of each token rather than frequency, without changing the base model. This method for the Portuguese essay scorer reduces the output layer vocabulary from about 152,000 lines to about 65,000 lines, significantly reducing the computational overhead of token splitting and unused vocabulary items.
Ant Group inclusionAI · Hugging Face new models3535inclusionAI released the UBTree fine-tuned version of Ling-3.0-tiny, Ling-3.0-tiny-UBTree-B16, with 183.1M parameters, combining a five-layer DFlash backbone and a bigram selector, achieving 2.6-3.5 times speed improvement on benchmarks like GSM8K and MATH-500. This model needs to be used in conjunction with the target model. The UBTree adaptation for the open-source version of SGLang has not yet been released.
NVIDIA Technical Blog3535The RFD4-Proteina can design octapeptide protein sequences, with over 2.5 billion combinations; 100-peptide sequences have possibilities exceeding the number of atoms in the observable universe, among which there are proteins that can be developed for new therapies; this model uses reward-guided reasoning to search for solutions to the protein design search problem.
GitHub Blog · AI & ML2222Three Java masters Vlad Mihalcea, Patrick Baumgartner, and Frank Delporte shared their experience using GitHub Copilot in projects such as Hypersistence Utils, Hypersistence Optimizer, and teaching slide migration.
OpenAI NewsOpenAI improves ChatGPT performance in low-resource programming languages, introduces new evaluation for software engineering, SQLite, and COBOL migration
OpenHands Blog6363Recursive Flint is released, a hybrid expert model with 4 billion parameters, designed for efficient local inference, activating only about 250 million parameters per token, achieving 125+ tokens/second on M4 Pro MacBook Pro, and reaching 150+ tokens/second on NVIDIA RTX Spark. OpenHands, as its first partner, supports the model's operation on local hardware, providing developers with an open-source coding model option.
Apple Machine Learning ResearchTEAL has for the first time efficiently applied active learning to the token space of ViT for low-budget pixel-level semantic segmentation, boosting mIoU by +24.36 on datasets like CamVid and Cityscapes with only one pixel labeled per image, and achieving full-supervised performance of 97% with 10 pixels labeled.
Tessl BlogTessl Blog stops line-by-line code review and starts verifying intent. The team's merged PR volume is four times that of last year and almost all AI-generated, making review a bottleneck. The team shifts to reviewing intent before code generation (acceptance criteria, constraints, and edge cases), and transforms repetitive review comments into machine-executable rules (invariants), thus shifting the focus of review from the code itself to decision-making and rule ownership.
The Qwen3-4B-deslop model eliminates common AI writing habits (such as dashes, 'not X, but Y' sentences, three-element lists, etc.) through fine-tuning, reducing the SLOP score from 121 to 12.5 on 40 unseen prompts, with 68% of responses reaching human writing level. This 4B model, after being quantized by Ollama, is only 2.5GB and can run locally while still maintaining a 30/30 fact question accuracy rate comparable to the original version.
Inception Labs BlogMercury 2 completes context compression tasks in just 150 seconds while maintaining the same quality as Sonnet 4.6, with a 5-fold speed increase and a 90% cost reduction. Augment Code uses it for enterprise-level coding AI agents, achieving an 82% reduction in compression latency and a 90% cost cut, and dynamically allocates tasks through the Prism model router, reducing total LLM expenditure by 30%.
Inception Labs BlogMercury Edit 2 is a new diffusion model (dLLM) launched by Inception Labs, used to predict the next step in code editing, with a 48% improvement in acceptance rate compared to the previous model, and more precise editing choices (improvement of 27%).
MiniMax NewsMiniMax M3.1 Flash is officially launched, with nearly a threefold increase in online inference speed and a 70% reduction in cost compared to the previous generation. The model performs excellently in benchmark tests for complex intelligent tasks (cowork), computer use, software development, and network security, ranking highly in the AutomationBench-AA for automated tasks, scoring 73.7% on DeepSWE v1.1, 61.7 on OSWorld V2, 29.6 on ALE, and 82% on CyberGym.
DeepSeek · Hugging Face new models5656DeepSeek-V4.1-Flash-1017 Released, Significantly Improving on Reasoning, Coding, Agent Tasks, and Visual Benchmarks, Reducing Execution Time and Output Length, Completing Tasks Faster and at Lower Cost. Scored 42.6 on HLE (No Tools) Benchmark, 51.8 on Terminal-Bench 4.0, 77.0 on DeepSWE v1.1, License is MIT.
Google ResearchGoogle Research releases the VERID framework, enabling AI agents to run in physics-based, safe, and resource-efficient experiments; demonstrates closed-loop intelligent agent scientific discovery on adapter and de novo protein design prototypes; through three rounds of design-test-iterate cycles, AI-driven rational and de novo design will increase confirmed target hit rates to 33.5% and 42.1% on myoglobin and α-thrombin, respectively.
OpenAI News3535Anant Kumar uses Codex for code review and to speed up follow-up.
vLLM BlogvLLM has built a unified CuTe DSL decoding path for Kimi K3 through intelligent agent-assisted development, achieving a batch 1 performance of 840 tokens/s on 16 NVIDIA GB300 GPUs, with a 1.5-2x increase in decoding throughput in the test configuration, and a 1.8x increase in AgentX workload P90 interactivity.
Baseten BlogMeta releases Muse Spark Open, an open-source weight model with 816B parameters designed for code agents, featuring 42.7B active parameters, a 1M token context window, and full-modal input (text, image, video, audio), performance comparable to cutting-edge lab models, using a compressed MoE architecture and hybrid global/local attention mechanisms to optimize inference efficiency, reducing model size to 540GB via NVFP4 checkpoints.
Ant Group inclusionAI · Hugging Face new modelsinclusionAI released the Ling-3-Flash-UBTree-B16 model, combining a five-layer DFlash backbone and a learned bigram selector, proposing multiple candidate continuations for Ling3-Flash for target model validation, with 482M parameters, achieving 3x+ E2E speed improvement on benchmarks like MMLU, and the UBTree adaptation is not yet open source, with SGLang integration support expected to be released soon.
Google DeepMindGemini 4 Argon demonstrates cutting-edge capabilities in coding, finance, law, and other fields, particularly enhancing cybersecurity defense functions. Its effectiveness has been validated through the Fairwind project and internal testing. The model performs excellently on benchmark tests like MMLU and employs a three-layer real-time protection mechanism (input filtering, internal introspection, continuous monitoring) and automated bias monitoring, significantly reducing the risk of harmful requests and violations. It successfully intercepts over 99% of external biosecurity jailbreak attempts while reducing shortcut behavior in coding tasks to 2%.
Claude · YouTubeClaude Code's cloud sessions run on cloud virtual machines, allowing offline continuous operation, supporting phone startup, and parallel processing. Suitable for scenarios requiring continuous operation or remote collaboration, supporting GitHub connection and environment configuration. This feature is included in the Pro, Max, Team plans, as well as enterprise advanced or Chat + Claude Code seats, counted towards plan limits with no additional compute costs.
Google DeepMindGemini 4 Argon ranked first on the Vals Index, surpassing other cutting-edge models in coding, finance, law, cybersecurity, and science with its 175B parameter size; its reasoning depth and accuracy make it a powerful tool for solving complex engineering problems, optimizing business processes, and accelerating scientific breakthroughs.
Artificial Analysis Articles7575Cognition released the FrontierCode 1.1 coding benchmark, evaluating whether AI model code is of sufficient quality to be merged by open-source maintainers, not only looking at test pass rates but also evaluating modification extent, scope, and file relevance.
Simon Willison6565Simon Willison interacted with GPT-6 Astra High via voice to build the Newsletters page of his blog, implementing a subscription list indexing feature, including weekly free Substack and monthly exclusive updates for sponsors. He used the Codex voice chat mode of the ChatGPT desktop application to complete the model, migration, view code, template, and database import functions in a local development environment and integrated the website search engine.
OpenAI News2222Oracle transforms expertise into fast, repeatable workflows in recruitment, engineering, and operations through ChatGPT Work and Codex, significantly improving efficiency.
Microsoft AI NewsMAI-Code-1.1-Flash optimizes with 3-bit quantization, halving memory requirements while retaining a 256K context window, with coding performance comparable to the full-precision version, SWE-Bench Verified and Terminal-Bench 2.1 scores comparable, with zero inference cost for local calls.
Latent Space7070Reflection announces the launch of Beam, a text-only 501B total parameters / 23B active parameters encoding, intelligent agent, and scientific work multimodal model using the MoE architecture, trained from scratch, and will open source all weights this month (Apache 2.0 license).
Together AI BlogTogether Link connects your team's existing tools to Together AI's top open-source models, reducing costs by over 50%. Supports Claude Code, Claude Desktop, Codex, OpenCode, and Pi, allowing switching without learning new tools, and can start using with a single command.
GitHub Blog · AI & MLDeveloper workflows are changing due to AI, with skills beyond coding—such as guiding AI, evaluating outputs, communicating trade-offs, and making technical decisions—becoming increasingly important. Three key skills developers should strengthen are learning to guide AI rather than just using it, not trusting AI's first answer and evaluating it, and leveraging AI to solve broader problems.
Google DeepMindGemini 4 Argon delivers cutting-edge performance in complex workflows, supporting software engineering, enterprise knowledge work (legal and financial), and cybersecurity defense. The model's token output limit is extended to the industry-leading 1M, and it leads in benchmarks such as DeepSWE v1.1 (77.9%), Vals Index (financial, coding, legal, tax), and AutomationBench (51.3%).
Anthropic ResearchAnthropic's Claude has successfully calculated the nine-loop amplitude angles for N=4 super Yang-Mills theory, breaking the previous two-loop calculation limit. The model is based on LLM technology and uses the CoT reasoning method.
Together AI BlogTogether AI released a guide on how to fine-tune the Jev classification model using the Qwen3.5 4B model and a Hugging Face dataset of 38,340 items, with a training cost of about $17 and a duration of about 25 minutes; after deployment, it can be used for inference via an API endpoint (e.g., ENDPOINT_NAME), suitable for scenarios such as intent recognition for e-commerce customers.
Xiaomi MiMo · Hugging Face new modelsXiaomi MiMo released the 9B parameter intelligent agent model MiMo-V2.6-Distill-Qwen-9B, which covers coding, general intelligent agent tasks, visual encoding, and cybersecurity fields by supervised fine-tuning of Qwen3.5-9B on data generated by MiMo.
Hamel Husain7575Use public benchmarks to evaluate code agents handling multiple tasks, such as SWE-bench and HumanEval; for narrow workflows, adopt product-specific evaluations; build private benchmarks containing organizational internal difficult tasks; identify issues through engineer reviews and error analysis and transform them into tests; Anthropic recommends starting with clearly specified tasks and stable environments with verifiable results, gradually adding extra checks like code quality or user interaction.
Cognition BlogDevin Desktop and CLI are now integrated with the Fusion architecture, improving efficiency by up to 39% compared to other code generation frameworks. Fusion allows users to select a cutting-edge model for planning and review ('lead') and a more cost-effective model for execution ('sidekick').
