Top Devstral Small 2 Alternatives

GLM-5.3 is Z.ai’s frontier coding model built to improve complex software engineering, long-horizon agent work, and advanced technical reasoning through scaled post-training. The model uses the same base model as GLM-5.2, with performance gains coming from additional post-training environments, more diverse tasks, and expanded compute on the existing training stack. Z.ai’s stack includes IndexShare for efficient long-context processing, SAO for reinforcement learning on long-horizon tasks, and slime for large-scale asynchronous post-training. GLM-5.3 is designed to perform better on work that resembles real engineering tasks rather than short coding exercises. Its training environments include production-style workflows where the model must diagnose bottlenecks, inspect documentation, use codebases, run experiments, implement changes, and produce measurable improvements. The model improves coding performance across public and private benchmarks, including Terminal Bench 3.0, DeepSWE, Agents’ Last Exam, and Z.ai Code Bench. GLM-5.3 also improves token efficiency, producing stronger agentic coding results than GLM-5.2 while using fewer output tokens in Z.ai’s internal evaluations. The model supports three reasoning effort levels, low, high, and max, and no longer supports disabling thinking. Z.ai recommends max reasoning effort for coding tasks, while applications using disabled thinking must migrate to enabled thinking before switching to GLM-5.3. The release also reports emergent cyber capabilities, including stronger vulnerability discovery and exploitation-chain reasoning, with open-weight release planned after safety evaluation and hardening. By combining scaled post-training, long-context infrastructure, long-horizon reinforcement learning, coding-agent workflows, benchmark improvements, reasoning controls, and ZCode integration, GLM-5.3 helps developers and researchers work on demanding coding and agentic tasks.

NVIDIA Nemotron 3.5 Bolt is a family of open artificial intelligence models developed to provide efficient reasoning, autonomous agent capabilities, and conversational intelligence on local computing hardware. Introduced in October 2026, the series includes Nemotron 3.5 Bolt 74B-A9B 24GB and Nemotron 3.5 Bolt E4B 4GB. The 74B-A9B variant uses a mixture-of-experts architecture containing 74 billion total parameters, with approximately 9 billion parameters activated during inference. Its native 2-bit quantization enables deployment within approximately 24GB of memory while preserving resources for other applications. The model is designed for agentic workflows that require reasoning, instruction following, tool interactions, and execution of multiple connected tasks. Developers can use Bolt 74B-A9B with local agent frameworks to conduct web research, generate software, analyze code, investigate programming errors, and perform debugging activities. NVIDIA reports strong results on agent benchmarks including PinchBench and BrowseComp, which evaluate autonomous task completion and information retrieval capabilities. The smaller Bolt E4B variant is designed for low-latency conversational AI, supporting local voice assistants, interactive game characters, and embedded interfaces. Its 4GB memory footprint makes it suitable for compatible GeForce RTX, NVIDIA Jetson, and DGX Spark systems that require efficient local inference. Running these models locally allows organizations to retain greater control over sensitive application data, reduce dependence on cloud inference, and integrate AI capabilities into existing software environments. Nemotron 3.5 Bolt 74B-A9B is available through Hugging Face with support for llama.cpp, Ollama, LM Studio, Unsloth, and vLLM, while Bolt E4B is scheduled for release on October 23, 2026.

DeepCoder, a fully open-source initiative for code reasoning and generation, has been created through a collaboration between the Agentica Project and Together AI. Built on the foundation of DeepSeek-R1-Distilled-Qwen-14B, it has been fine-tuned using distributed reinforcement learning techniques, achieving an impressive accuracy of 60.6% on LiveCodeBench, which represents an 8% improvement compared to its predecessor. This remarkable performance positions it competitively alongside proprietary models such as o3-mini (2025-01-031 Low) and o1, all while operating with a streamlined 14 billion parameters. The training process was intensive, lasting 2.5 weeks on a fleet of 32 H100 GPUs and utilizing a meticulously curated dataset comprising around 24,000 coding challenges obtained from reliable sources such as TACO-Verified, PrimeIntellect SYNTHETIC-1, and submissions to LiveCodeBench. Each coding challenge was required to include a valid solution paired with at least five unit tests to ensure robustness during the reinforcement learning phase. Additionally, DeepCoder employs innovative methods like iterative context lengthening and overlong filtering to effectively handle long-range contextual dependencies, allowing it to tackle complex coding tasks with proficiency. This distinctive approach not only enhances DeepCoder's accuracy and reliability in code generation but also positions it as a significant player in the landscape of code generation models. As a result, developers can rely on its capabilities for diverse programming challenges.

Laguna S 2.1 represents a state-of-the-art open weight coding model that focuses on the completion of long-term projects and demonstrates exceptional reasoning abilities. With a Mixture-of-Experts architecture comprising 118 billion parameters, it engages 8 billion parameters per token and supports a context window of up to one million tokens in both cognitive and non-cognitive modes. The model’s optimized active size enables it to execute complex tasks on local systems while remaining competitive with much larger models across a variety of benchmarks, such as terminal usage, software development, codebase question answering, and tool application. Built for durability, Laguna S 2.1 is adept at addressing demanding challenges with an emphasis on thorough verification and a willingness to backtrack when necessary, rather than hastily claiming victory. In real-world scenarios, it has successfully engineered a browser rendering engine from the ground up, improved an agent harness for faster execution and lower memory requirements, and conducted comprehensive mathematical investigations using the tools available in its environment, showcasing its adaptability and proficiency. This remarkable array of capabilities positions Laguna S 2.1 as an invaluable asset for developers in search of cutting-edge solutions, making it a top choice in the ever-evolving landscape of coding models.

Laguna XS.2 stands out as Poolside's groundbreaking open-weight coding model, noted for being the lightest and fastest in the Laguna lineup. Equipped with a staggering 33 billion parameters organized in a Mixture of Experts structure, of which 3 billion are active, this model has undergone extensive training in-house utilizing 30 trillion tokens. As the most recent generation model available to the public, it features a second-generation architecture and represents Poolside's first open-weight release, benefiting from lessons learned during the Laguna M.1 training process, which utilized synthetic data and reinforcement learning. Tailored specifically to optimize agentic coding workflows, Laguna XS.2 is exceptional in coding, acting, and rapid iteration, particularly within Poolside's coding agent ecosystem. This model is especially beneficial for developers and teams in need of a lightweight and efficient coding solution, as opposed to more complex frontier systems. Released under the flexible Apache 2.0 license, it enables the community to evaluate, refine, quantize, and build upon its weights, fostering an environment of collaborative development. Ultimately, Laguna XS.2 not only serves as a powerful tool for agentic coding but also promotes creativity and experimentation among its users, allowing for a diverse range of applications and enhancements.

DeepSWE represents a groundbreaking advancement in open-source coding agents, harnessing the Qwen3-32B foundation model trained exclusively through reinforcement learning (RL) without the aid of supervised fine-tuning or proprietary model distillation. Developed using rLLM, which is Agentica's open-source RL framework tailored for language-driven agents, DeepSWE functions effectively within a simulated development environment provided by the R2E-Gym framework. This setup equips it with a range of tools, such as a file editor, search functions, shell execution, and submission capabilities, allowing the agent to adeptly navigate extensive codebases, modify multiple files, compile code, execute tests, and iteratively generate patches or fulfill intricate engineering tasks. In addition to mere code generation, DeepSWE exhibits sophisticated emergent behaviors; when confronted with bugs or feature requests, it engages in critical reasoning regarding edge cases, searches for existing tests in the codebase, proposes patches, creates additional tests to avert regressions, and adapts its cognitive strategies based on the specific challenges presented. This remarkable adaptability and efficiency position DeepSWE as a formidable asset in the software development landscape, empowering developers to tackle complex projects with greater ease and confidence. Its ability to learn from each interaction further enhances its performance, ensuring continuous improvement over time.

North Mini Code marks the launch of Cohere's innovative agentic coding model, specifically designed for developers, and represents the initial offering in its next generation of advanced models. This compact and effective open-source solution is tailored for the independent developer community, providing exceptional software development capabilities without requiring extensive hardware resources. Utilizing a mixture-of-experts architecture, it features a total of 30 billion parameters, with 3 billion actively engaged, delivering powerful agentic coding functionalities in a streamlined format. The model is meticulously optimized for a variety of tasks, including code generation, agentic software engineering, and terminal operations, boasting an impressive context length of 256K and a maximum generation capacity of 64K. It is crafted with real-world developer practices in mind, allowing for the management of sub-agents, architecture mapping, code reviews, and supporting coding agents in overcoming complex software challenges. By integrating these capabilities, developers can significantly boost their productivity and efficiency in software development projects, making it an invaluable tool in their arsenal. As a result, North Mini Code not only facilitates better coding practices but also fosters a collaborative environment for developers to thrive.

The Nemotron 3 Nano distinguishes itself as the smallest model in NVIDIA's Nemotron 3 series, tailored specifically for agentic AI applications that necessitate strong reasoning and conversational capabilities while ensuring economical inference costs. This innovative hybrid Mamba-Transformer Mixture-of-Experts model is equipped with 3.2 billion active parameters and expands to 3.6 billion when accounting for embeddings, culminating in an impressive total of 31.6 billion parameters. NVIDIA claims that this model achieves superior accuracy compared to its predecessor, the Nemotron 2 Nano, while also operating with less than half of the parameters during each forward pass, thereby boosting efficiency without sacrificing performance. Additionally, it reportedly outperforms both GPT-OSS-20B and Qwen3-30B-A3B-Thinking-2507 across a range of commonly used benchmarks. With an input capacity of 8K and an output limit of 16K utilizing a single H200, the model realizes an inference throughput that is 3.3 times higher than that of Qwen3-30B-A3B and 2.2 times that of GPT-OSS-20B. Furthermore, the Nemotron 3 Nano can manage context lengths of up to 1 million tokens, reinforcing its dominance over GPT-OSS-20B and Qwen3-30B-A3B-Instruct-2507. This extraordinary amalgamation of capabilities not only enhances its precision and efficiency but also positions the Nemotron 3 Nano as a premier option for cutting-edge AI endeavors that require top-tier performance. As the demand for advanced AI solutions grows, the relevance of such models will likely continue to expand.

Devstral represents a joint initiative by Mistral AI and All Hands AI, creating an open-source large language model designed explicitly for the field of software engineering. This innovative model exhibits exceptional skill in navigating complex codebases, efficiently managing edits across multiple files, and tackling real-world issues, achieving an impressive 46.8% score on the SWE-Bench Verified benchmark, which positions it ahead of all other open-source models. Built upon the foundation of Mistral-Small-3.1, Devstral features a vast context window that accommodates up to 128,000 tokens. It is optimized for peak performance on advanced hardware configurations, such as Macs with 32GB of RAM or Nvidia RTX 4090 GPUs, and is compatible with several inference frameworks, including vLLM, Transformers, and Ollama. Released under the Apache 2.0 license, Devstral is readily available on various platforms, including Hugging Face, Ollama, Kaggle, Unsloth, and LM Studio, enabling developers to effortlessly incorporate its features into their applications. This model not only boosts efficiency for software engineers but also acts as a crucial tool for anyone engaged in coding tasks, thereby broadening its utility and appeal across the tech community. Furthermore, its open-source nature encourages continuous improvement and collaboration among developers worldwide.

Devstral 2 is an innovative, open-source AI model tailored for software engineering, transcending simple code suggestions to fully understand and manipulate entire codebases; this advanced functionality enables it to execute tasks such as multi-file edits, bug fixes, refactoring, managing dependencies, and generating code that is aware of its context. The suite includes a powerful 123-billion-parameter model alongside a streamlined 24-billion-parameter variant called “Devstral Small 2,” offering flexibility for teams; the larger model excels in handling intricate coding tasks that necessitate a deep contextual understanding, whereas the smaller model is optimized for use on less robust hardware. With a remarkable context window capable of processing up to 256 K tokens, Devstral 2 is adept at analyzing extensive repositories, tracking project histories, and maintaining a comprehensive understanding of large files, which is especially advantageous for addressing the challenges of real-world software projects. Additionally, the command-line interface (CLI) further enhances the model’s functionality by monitoring project metadata, Git statuses, and directory structures, thereby enriching the AI’s context and making “vibe-coding” even more impactful. This powerful blend of features solidifies Devstral 2's role as a revolutionary tool within the software development ecosystem, offering unprecedented support for engineers. As the landscape of software engineering continues to evolve, tools like Devstral 2 promise to redefine the way developers approach coding tasks.

GPT-5.2-Codex is OpenAI’s most capable agentic coding model, engineered for professional software engineering and cybersecurity use cases. It builds on the strengths of GPT-5.2 while introducing optimizations for long-running coding sessions. The model excels at maintaining context across extended workflows using native context compaction. GPT-5.2-Codex performs reliably in large repositories and complex project structures. It achieves state-of-the-art results on SWE-Bench Pro and Terminal-Bench 2.0, reflecting strong real-world coding performance. Native Windows support improves reliability for cross-platform development. Enhanced vision capabilities allow the model to interpret design mocks, diagrams, and screenshots. GPT-5.2-Codex supports iterative development even when plans change or attempts fail. The model also shows substantial gains in defensive cybersecurity tasks. It can assist with vulnerability discovery and secure software development workflows. Additional safeguards are built in to address dual-use risks. GPT-5.2-Codex advances the frontier of agentic software engineering.

Mistral Small 4 is a powerful open-source AI model introduced by Mistral AI to deliver advanced reasoning, multimodal understanding, and coding capabilities in a single system. The model represents the latest evolution in the Mistral Small family and consolidates multiple specialized AI technologies into one unified architecture. It integrates the reasoning capabilities of Magistral, the multimodal functionality of Pixtral, and the coding intelligence of Devstral. This design allows the model to handle tasks ranging from conversational assistance and research analysis to software development and visual data processing. Mistral Small 4 supports both text and image inputs, enabling applications such as document parsing, visual analysis, and interactive AI systems. Its mixture-of-experts architecture includes 128 experts with a small subset activated per token, allowing efficient resource usage while maintaining strong performance. The model also introduces a configurable reasoning effort parameter that allows developers to control the balance between speed and analytical depth. A large 256k context window enables it to process lengthy conversations, documents, and complex reasoning workflows. Performance optimizations significantly reduce latency and increase throughput compared with previous versions of the model. The system is designed for deployment across various environments, including cloud infrastructure, enterprise systems, and research environments. Developers can access the model through platforms such as Hugging Face, Transformers, and optimized inference frameworks. Released under the Apache 2.0 open-source license, Mistral Small 4 allows organizations to customize, fine-tune, and deploy AI solutions tailored to their specific needs. By combining reasoning, multimodal processing, and coding intelligence in one model, Mistral Small 4 simplifies AI integration for modern applications.

Stable LM signifies a notable progression in the language model domain, building upon prior open-source experiences, especially through collaboration with EleutherAI, a nonprofit research group. This evolution has included the creation of prominent models like GPT-J, GPT-NeoX, and the Pythia suite, all trained on The Pile open-source dataset, with several recent models such as Cerebras-GPT and Dolly-2 taking cues from this foundational work. In contrast to earlier models, Stable LM utilizes a groundbreaking dataset that is three times as extensive as The Pile, comprising an impressive 1.5 trillion tokens. More details regarding this dataset will be disclosed soon. The vast scale of this dataset allows Stable LM to perform exceptionally well in conversational and programming tasks, even though it has a relatively compact parameter size of 3 to 7 billion compared to larger models like GPT-3, which features 175 billion parameters. Built for adaptability, Stable LM 3B is a streamlined model designed to operate efficiently on portable devices, including laptops and mobile gadgets, which excites us about its potential for practical usage and portability. This innovation has the potential to bridge the gap for users seeking advanced language capabilities in accessible formats, thus broadening the reach and impact of language technologies. Overall, the launch of Stable LM represents a crucial advancement toward developing more efficient and widely available language models for diverse users.

Inkling-Small is an efficient multimodal AI model built to deliver strong reasoning and coding performance at a fraction of Inkling’s size. It is a Mixture-of-Experts transformer with 276 billion total parameters and 12 billion active parameters. The model was trained on NVIDIA GB300 NVL72 systems and is designed to combine high capability with more efficient inference. Inkling-Small supports native reasoning across text, images, and audio, allowing it to work across multimodal tasks without relying on separate encoders. Its context window supports up to one million tokens, making it useful for long-form reasoning, large-scale code understanding, document analysis, and agentic workflows. Users can adjust reasoning effort from minimal to extra high depending on whether they need faster responses or deeper computation. The model’s training process includes improved pre-training data, post-training with on-policy distillation from Inkling, and extended agentic coding reinforcement learning. These techniques helped Inkling-Small outperform its larger counterpart on reasoning and coding benchmarks. The model performs well in coding and tool-use harnesses and exceeds 80% on SWE-bench Verified. Its encoder-free architecture processes audio as dMel spectrograms and images as 40-by-40-pixel patches alongside text tokens. By combining efficient MoE design, one-million-token context, adjustable reasoning effort, multimodal processing, coding strength, and tool-use performance, Inkling-Small is designed for developers and teams that need capable AI with lower active compute requirements.

Claude Opus 4.5 represents a major leap in Anthropic’s model development, delivering breakthrough performance across coding, research, mathematics, reasoning, and agentic tasks. The model consistently surpasses competitors on SWE-bench Verified, SWE-bench Multilingual, Aider Polyglot, BrowseComp-Plus, and other cutting-edge evaluations, demonstrating mastery across multiple programming languages and multi-turn, real-world workflows. Early users were struck by its ability to handle subtle trade-offs, interpret ambiguous instructions, and produce creative solutions—such as navigating airline booking rules by reasoning through policy loopholes. Alongside capability gains, Opus 4.5 is Anthropic’s safest and most robustly aligned model, showing industry-leading resistance to strong prompt-injection attacks and lower rates of concerning behavior. Developers benefit from major upgrades to the Claude API, including effort controls that balance speed versus capability, improved context efficiency, and longer-running agentic processes with richer memory. The platform also strengthens multi-agent coordination, enabling Opus 4.5 to manage subagents for complex, multi-step research and engineering tasks. Claude Code receives new enhancements like Plan Mode improvements, parallel local and remote sessions, and better GitHub research automation. Consumer apps gain better context handling, expanded Chrome integration, and broader access to Claude for Excel. Enterprise and premium users see increased usage limits and more flexible access to Opus-level performance. Altogether, Claude Opus 4.5 showcases what the next generation of AI can accomplish—faster work, deeper reasoning, safer operation, and richer support for modern development and productivity workflows.

Qwen3-Coder is a multifaceted coding model available in different sizes, prominently showcasing the 480B-parameter Mixture-of-Experts variant with 35B active parameters, which adeptly manages 256K-token contexts that can be scaled up to 1 million tokens. It demonstrates remarkable performance comparable to Claude Sonnet 4, having been pre-trained on a staggering 7.5 trillion tokens, with 70% of that data comprising code, and it employs synthetic data fine-tuned through Qwen2.5-Coder to bolster both coding proficiency and overall effectiveness. Additionally, the model utilizes advanced post-training techniques that incorporate substantial, execution-guided reinforcement learning, enabling it to generate a wide array of test cases across 20,000 parallel environments, thus excelling in multi-turn software engineering tasks like SWE-Bench Verified without requiring test-time scaling. Beyond the model itself, the open-source Qwen Code CLI, inspired by Gemini Code, equips users to implement Qwen3-Coder within dynamic workflows by utilizing customized prompts and function calling protocols while ensuring seamless integration with Node.js, OpenAI SDKs, and environment variables. This robust ecosystem not only aids developers in enhancing their coding projects efficiently but also fosters innovation by providing tools that adapt to various programming needs. Ultimately, Qwen3-Coder stands out as a powerful resource for developers seeking to improve their software development processes.

GLM-4.1V represents a cutting-edge vision-language model that provides a powerful and efficient multimodal ability for interpreting and reasoning through different types of media, such as images, text, and documents. The 9-billion-parameter variant, referred to as GLM-4.1V-9B-Thinking, is built on the GLM-4-9B foundation and has been refined using a distinctive training method called Reinforcement Learning with Curriculum Sampling (RLCS). With a context window that accommodates 64k tokens, this model can handle high-resolution inputs, supporting images with a resolution of up to 4K and any aspect ratio, enabling it to perform complex tasks like optical character recognition, image captioning, chart and document parsing, video analysis, scene understanding, and GUI-agent workflows, which include interpreting screenshots and identifying UI components. In benchmark evaluations at the 10 B-parameter scale, GLM-4.1V-9B-Thinking achieved remarkable results, securing the top performance in 23 of the 28 tasks assessed. These advancements mark a significant progression in the fusion of visual and textual information, establishing a new benchmark for multimodal models across a variety of applications, and indicating the potential for future innovations in this field. This model not only enhances existing workflows but also opens up new possibilities for applications in diverse domains.

Upstage has introduced Solar Pro 2, a state-of-the-art large language model engineered for frontier-scale applications, adept at handling complex tasks and workflows across multiple domains such as finance, healthcare, and legal fields. This model features a streamlined architecture with 31 billion parameters, delivering outstanding multilingual support, particularly excelling in Korean, where it outperforms even larger models on significant benchmarks like Ko-MMLU, Hae-Rae, and Ko-IFEval, while also maintaining solid performance in English and Japanese. Beyond its impressive language understanding and generation skills, Solar Pro 2 integrates an advanced Reasoning Mode that greatly improves the precision of multi-step tasks across various challenges, ranging from general reasoning tests (MMLU, MMLU-Pro, HumanEval) to complex mathematical problems (Math500, AIME) and software engineering assessments (SWE-Bench Agentless), achieving problem-solving efficiencies that rival or exceed those of models with twice the number of parameters. Additionally, its superior tool-use capabilities enable the model to interact effectively with external APIs and datasets, enhancing its relevance in practical applications. This groundbreaking architecture not only showcases remarkable adaptability but also establishes Solar Pro 2 as a significant contender in the rapidly advancing field of AI technologies, paving the way for future innovations. As the demand for advanced AI solutions continues to grow, Solar Pro 2 is poised to meet the challenges of various industries head-on.

Reka Flash 3 stands as a state-of-the-art multimodal AI model, boasting 21 billion parameters and developed by Reka AI, to excel in diverse tasks such as engaging in general conversations, coding, adhering to instructions, and executing various functions. This innovative model skillfully processes and interprets a wide range of inputs, which includes text, images, video, and audio, making it a compact yet versatile solution fit for numerous applications. Constructed from the ground up, Reka Flash 3 was trained on a diverse collection of datasets that include both publicly accessible and synthetic data, undergoing a thorough instruction tuning process with carefully selected high-quality information to refine its performance. The concluding stage of its training leveraged reinforcement learning techniques, specifically the REINFORCE Leave One-Out (RLOO) method, which integrated both model-driven and rule-oriented rewards to enhance its reasoning capabilities significantly. With a remarkable context length of 32,000 tokens, Reka Flash 3 effectively competes against proprietary models such as OpenAI's o1-mini, making it highly suitable for applications that demand low latency or on-device processing. Operating at full precision, the model requires a memory footprint of 39GB (fp16), but this can be optimized down to just 11GB through 4-bit quantization, showcasing its flexibility across various deployment environments. Furthermore, Reka Flash 3's advanced features ensure that it can adapt to a wide array of user requirements, thereby reinforcing its position as a leader in the realm of multimodal AI technology. This advancement not only highlights the progress made in AI but also opens doors to new possibilities for innovation across different sectors.

SWE-2 Flash is an AI coding and software engineering model developed by Cognition to deliver high-speed programming capabilities with lower inference costs than larger frontier models. The model uses a 276-billion-parameter mixture-of-experts architecture with approximately 12 billion active parameters, based on Thinking Machines' Inkling-Small foundation model. Cognition enhances the underlying model through a dedicated mid-training process followed by large-scale reinforcement learning focused on software engineering and general reasoning tasks. Its training incorporates several trillion tokens of carefully selected technical material, including source code, scientific publications, articles, and other datasets relevant to complex problem-solving. SWE-2 Flash supports automated code generation, software maintenance, code reviews, feature development, documentation updates, and other tasks performed by autonomous development agents. The model is optimized for efficient inference and generates approximately 200 tokens per second, compared with approximately 60 tokens per second for the larger SWE-2 model. Cognition reports that SWE-2 Flash operates at one-fifth the cost of SWE-2 while maintaining competitive performance across several software engineering evaluations. Its published benchmark results include 24.0% on FrontierCode 2.0, 45.5% on FrontierCode 1.1 Main, 60.1% on DeepSWE 1.1, and 89.9% on SWE-Bench Pro v2 Hard. The model also demonstrates cybersecurity-related capabilities, achieving 34.8% on ExploitBench compared with 11.3% for its original foundation model. Developers can access SWE-2 Flash through Devin Desktop and CLI for interactive programming assistance and agent-driven engineering workflows. Cognition is developing a secure program for sharing model weights with approved industry partners and research organizations, potentially enabling fine-tuning and deployment on private or on-premises infrastructure.

Composer 2 is a cutting-edge AI coding model integrated into Cursor, designed to deliver frontier-level programming intelligence with strong efficiency and cost optimization. It is built on advanced pretraining and reinforcement learning techniques, enabling it to handle complex, long-horizon coding tasks that require hundreds of steps and decisions. The model demonstrates significant improvements across key benchmarks, including Terminal-Bench and SWE-bench Multilingual, highlighting its ability to perform in real-world development scenarios. Composer 2 excels at generating accurate, high-quality code while maintaining fast processing speeds, making it ideal for demanding workflows. Its architecture allows it to break down complex problems, plan solutions, and execute them effectively across different programming contexts. The model is available at competitive pricing, making advanced AI coding capabilities more accessible to developers. It also offers a faster variant that maintains the same intelligence while delivering improved speed for rapid execution tasks. Integrated within the Cursor environment, it enables seamless interaction with coding workflows and tools. Composer 2 is designed to support a wide range of use cases, from debugging and refactoring to building complex applications. Its ability to handle multi-step reasoning makes it especially valuable for large-scale projects. By combining performance, speed, and affordability, it sets a new standard for AI-assisted development. Overall, Composer 2 empowers developers to write better code faster and more efficiently.

We are excited to unveil Mistral NeMo, our latest and most sophisticated small model, boasting an impressive 12 billion parameters and a vast context length of 128,000 tokens, all available under the Apache 2.0 license. In collaboration with NVIDIA, Mistral NeMo stands out in its category for its exceptional reasoning capabilities, extensive world knowledge, and coding skills. Its architecture adheres to established industry standards, ensuring it is user-friendly and serves as a smooth transition for those currently using Mistral 7B. To encourage adoption by researchers and businesses alike, we are providing both pre-trained base models and instruction-tuned checkpoints, all under the Apache license. A remarkable feature of Mistral NeMo is its quantization awareness, which enables FP8 inference while maintaining high performance levels. Additionally, the model is well-suited for a range of global applications, showcasing its ability in function calling and offering a significant context window. When benchmarked against Mistral 7B, Mistral NeMo demonstrates a marked improvement in comprehending and executing intricate instructions, highlighting its advanced reasoning abilities and capacity to handle complex multi-turn dialogues. Furthermore, its design not only enhances its performance but also positions it as a formidable option for multi-lingual tasks, ensuring it meets the diverse needs of various use cases while paving the way for future innovations.

Magistral marks the first language model family launched by Mistral AI, focusing on enhanced reasoning abilities and available in two distinct versions: Magistral Small, which is a 24 billion parameter model with open weights under the Apache 2.0 license and can be found on Hugging Face, and Magistral Medium, a more advanced version designed for enterprise use, accessible through Mistral's API, the Le Chat platform, and several leading cloud marketplaces. Tailored for specific sectors, this model excels at transparent, multilingual reasoning across a variety of tasks, including mathematics, physics, structured calculations, programmatic logic, decision trees, and rule-based systems, producing outputs that maintain a coherent thought process in the language preferred by the user, enabling easy tracking and validation of results. The launch of this model signifies a notable shift towards compact yet highly efficient AI reasoning capabilities that are easily interpretable. Presently, Magistral Medium is available in preview on platforms such as Le Chat, the API, SageMaker, WatsonX, Azure AI, and Google Cloud Marketplace. Its architecture is specifically designed for general-purpose tasks that require prolonged cognitive engagement and enhanced precision in comparison to conventional non-reasoning language models. The arrival of Magistral is a landmark achievement that showcases the ongoing evolution towards more sophisticated reasoning in artificial intelligence applications, setting new standards for performance and usability. As more organizations explore these capabilities, the potential impact of Magistral on various industries could be profound.

MAI-Cyber-1-Flash is a sophisticated security framework from Microsoft AI, specifically designed to identify vulnerabilities within complex code. It is part of the MAI-Thinking-1 family and has been developed using high-quality data, being fully integrated into MDASH, which is Microsoft's extensive platform for vulnerability detection and resolution through a network of agents. MDASH utilizes more than 100 finely-tuned agents and advanced models to efficiently find, verify, and fix software vulnerabilities, while MAI-Cyber-1-Flash is capable of handling up to 90% of the associated tasks. For more intricate challenges, larger models like GPT-5.4 can be utilized, providing a well-calibrated multi-model strategy that accurately assigns the best model for each task. The synergy between MDASH and MAI-Cyber-1-Flash has led to a remarkable achievement of 96% performance on CyberGym, outperforming other competitors such as Mythos, Gemini, and various GPT-based solutions in their capability to analyze large codebases for vulnerability identification. These technological advancements not only enhance security measures but also represent a significant progression in maintaining the safety and reliability of software systems amidst an increasingly intricate digital environment. The ongoing collaboration between these innovative technologies promises to further revolutionize the field of cybersecurity.

Mistral 3 marks the latest development in the realm of open-weight AI models created by Mistral AI, featuring a wide array of options ranging from small, edge-optimized variants to a prominent large-scale multimodal model. Among this selection are three streamlined “Ministral 3” models, equipped with 3 billion, 8 billion, and 14 billion parameters, specifically designed for use on resource-constrained devices like laptops, drones, and various edge devices. In addition, the powerful “Mistral Large 3” serves as a sparse mixture-of-experts model, featuring an impressive total of 675 billion parameters, with 41 billion actively utilized. These models are adept at managing multimodal and multilingual tasks, excelling in areas such as text analysis and image understanding, and have demonstrated remarkable capabilities in responding to general inquiries, handling multilingual conversations, and processing multimodal inputs. Moreover, both the base and instruction-tuned variants are offered under the Apache 2.0 license, which promotes significant customization and integration into a range of enterprise and open-source projects. This approach not only enhances flexibility in usage but also sparks innovation and fosters collaboration among developers and organizations, ultimately driving advancements in AI technology.

LFM3 showcases Liquid AI's collection of versatile, open-weight foundational models tailored for a diverse range of applications, spanning text, vision, audio, and multimodal functionalities. This array includes specialized versions for generating text, processing vision-language tasks, handling audio, addressing omni tasks, and facilitating decision-making processes. Developers can choose to download the model weights for customization through fine-tuning and deploy them on local edge devices, or they may opt for Liquid AI’s hosted API for seamless access. The potential uses for these models are extensive, including conversational agents, invoking tools, transcribing and synthesizing speech, interpreting images and videos, extracting and classifying documents, routing information, and enabling robotics capabilities. LFM3 is designed to meet the needs of developers, researchers, and organizations aiming to build customized AI solutions that prioritize adaptable deployment options, including support for on-premises inference. Additionally, these open-weight models are governed by Liquid AI’s LFM Open License, which stipulates the eligibility for commercial use and its specific terms based on the annual revenue of the organization. This level of flexibility empowers users to precisely customize their AI implementations to suit their specific requirements, thereby significantly enhancing the overall functionality and effectiveness of the models. Moreover, with the rapid advancements in AI, having such adaptable tools is crucial for staying competitive in the ever-evolving technological landscape.

Qwen3.8-27B is an open-weights 27B-class model connected to Alibaba’s Qwen3.8 release, built for developers, researchers, and AI teams that need a capable but more deployable model size. Alibaba’s Qwen3.8 launch described the broader model family as optimized for coding and cowork scenarios, including software development, document processing, data analysis, and professional workflows. Reports state that Alibaba planned to open-source Qwen3.8-Max alongside Qwen3.8-27B, expanding access for developers and researchers. Qwen3.8-27B gives builders a smaller alternative to the 2.4T-parameter Qwen3.8-Max model, which third-party coverage describes as Qwen’s first Max-scale model planned for open weights. The model is well suited for coding assistance, local development, agent testing, workflow automation, data analysis, document understanding, and private AI experimentation. QwenCloud documentation lists Qwen3.8-Max as supporting a 1M context window, thinking, function calling, built-in tools, and structured output, showing the broader Qwen3.8 generation’s focus on advanced agent and application workflows. Qwen3.8-27B is especially useful for teams that want Qwen-family capabilities without the infrastructure demands of Max-scale deployment. Community posts around the release point to active interest in Hugging Face, Unsloth GGUF, Ollama, and local inference use cases. Third-party coverage also notes practical hardware discussions around quantized Qwen3.8-27B deployment, including claims that 4-bit variants can fit more easily on consumer or workstation GPUs. The model can be positioned for organizations that need open AI infrastructure, coding agents, local model evaluation, private deployments, and cost-controlled experimentation. By combining open-weight access, a practical 27B model size, Qwen3.8-era performance ambitions, coding-oriented workflows, and local deployment interest, Qwen3.8-27B gives developers a flexible foundation for building AI products and agents.

GLM-4.7 is an advanced AI model engineered to push the boundaries of coding, reasoning, and agent-based workflows. It delivers clear performance gains across software engineering benchmarks, terminal automation, and multilingual coding tasks. GLM-4.7 enhances stability through interleaved, preserved, and turn-level thinking, enabling better long-horizon task execution. The model is optimized for use in modern coding agents, making it suitable for real-world development environments. GLM-4.7 also improves creative and frontend output, generating cleaner user interfaces and more visually accurate slides. Its tool-using abilities have been significantly strengthened, allowing it to interact with browsers, APIs, and automation systems more reliably. Advanced reasoning improvements enable better performance on mathematical and logic-heavy tasks. GLM-4.7 supports flexible deployment, including cloud APIs and local inference. The model is compatible with popular inference frameworks such as vLLM and SGLang. Developers can integrate GLM-4.7 into existing workflows with minimal configuration changes. Its pricing model offers high performance at a fraction of comparable coding models. GLM-4.7 is designed to feel like a dependable coding partner rather than just a benchmark-optimized model.

GLM-4.5V-Flash is an open-source vision-language model designed to seamlessly integrate powerful multimodal capabilities into a streamlined and deployable format. This versatile model supports a variety of input types including images, videos, documents, and graphical user interfaces, enabling it to perform numerous functions such as scene comprehension, chart and document analysis, screen reading, and image evaluation. Unlike larger models, GLM-4.5V-Flash boasts a smaller size yet retains crucial features typical of visual language models, including visual reasoning, video analysis, GUI task management, and intricate document parsing. Its application within "GUI agent" frameworks allows the model to analyze screenshots or desktop captures, recognize icons or UI elements, and facilitate both automated desktop and web activities. Although it may not reach the performance levels of the most extensive models, GLM-4.5V-Flash offers remarkable adaptability for real-world multimodal tasks where efficiency, lower resource demands, and broad modality support are vital. Ultimately, its innovative design empowers users to leverage sophisticated capabilities while ensuring optimal speed and easy access for various applications. This combination makes it an appealing choice for developers seeking to implement multimodal solutions without the overhead of larger systems.

Muse Glimmer is a cutting-edge model boasting 30 billion parameters, crafted by Meta Superintelligence Labs, specifically optimized for seamless local agent functionality. Its streamlined architecture enables operation on standard Mac or PC systems with a single consumer GPU, making it suitable for a range of applications, including local agent management, programming tasks, function invocation, and evaluations within LLM-as-a-judge scenarios, all without needing cloud services or an internet connection. This groundbreaking model features sophisticated abilities like long-horizon execution, precise tool invocation, multimodal understanding, expanded memory for contextual awareness, and proficient instruction adherence. It excels in performing comprehensive tasks as an agent, adeptly navigates complex multi-step reasoning across extensive workflows, and can recover effectively from unexpected tool interactions. Additionally, it interprets interleaved text and images through a specialized perception encoder tailored for analyzing screenshots, graphs, and various document types. Beyond its primary functions, Muse Glimmer is designed to work harmoniously with OpenClaw and other orchestration frameworks, allowing for customizable reasoning capabilities and has been trained on a rich dataset that spans over 100 languages. The adaptability of this model not only enhances its effectiveness across different fields but also positions it as a significant asset in the evolving landscape of AI applications. Its innovative features and user-friendly deployment make it a versatile choice for professionals seeking to leverage AI for complex problem-solving.