Grok 4.6 is the latest version of xAI's Grok AI model, officially released on August 12, 2026. Musk flagged it publicly on August 20. According to xAI, the update focuses on three areas: upgraded multimodal understanding (text, images, and video), long-running agentic work, and the introduction of Grok Bot — a persistent AI agent that runs on a cloud computer rather than in a single session window. ### What does Grok Bot actually do? [...] According to Musk, image and video understanding in Grok 4.6 has undergone significant upgrades. The model can now handle visual inputs alongside natural language with greater accuracy — meaning you can share an image or video clip and expect more precise, contextually aware analysis in return. For practical use cases, this matters for anything from reading charts and diagrams to interpreting real-world footage. ### Is this relevant to Tesla owners specifically? [...] Elon Musk announced Grok 4.6 on August 20, pointing to a significant capability jump for xAI's conversational AI — including the ability to generate videos that explain complex concepts. The update, which officially launched on August 12 according to xAI, pushes Grok further into multimodal and agentic territory, moving it well beyond a simple chat interface. Elon Musk announces Grok 4.6 on X ### What is Grok 4.6, and when did it launch?
Daily briefing
Model news
Daily headlines around the freshest and most-listed models in the catalog, fetched via Tavily. Also available as RSS or JSON.
Xinmei Shen Alibaba Group Holding’s new lightweight AI model Qwen3.8-27B has matched much larger near-frontier rivals while being able to run on everyday hardware, impressing developers as local AI gains momentum. The Qwen3.8-27B, a small model with 27 billion parameters, performed on par with OpenAI’s GPT-5.6 Luna, which was billed as the most cost-efficient model in the US lab’s latest flagship series, benchmark firm Artificial Analysis said on Monday. [...] Artificial intelligence TechTech Trends # Alibaba’s lightweight Qwen model takes on larger AI systems from OpenAI, DeepSeek, Zhipu Chinese tech giant’s latest offering performed on par with OpenAI’s GPT-5.6 Luna and nearly matched DeepSeek, Zhipu’s open-weight models 2-MIN READ2-MIN Chinese tech titan Alibaba Group Holding’s new lightweight AI model Qwen3.8-27B has matched much larger rivals including OpenAI, DeepSeek and Zhipu. Photo: Shutterstock Xinmei Shen [...] The new findings came days after Alibaba released Qwen3.8-27B’s model weights – the underlying parameters that encode its intelligence – last Friday. On Artificial Analysis’ Agentic Index, which measures models’ performance in AI agent-focused workflows, Alibaba’s small model outperformed GPT-5.6 series’ mid-tier model Terra and Anthropic’s powerful Claude Opus 4.8 released in May. Advertisement Select Voice Select Speed
DeepSeek shipped the general-availability “0813” build of V4 Pro on August 13, 2026, replacing the April preview that had been running behind the same model ID. It is a mixture-of-experts model with roughly 1.6 trillion total parameters and about 49 billion active per token, built on a DeepSeekV4ForCausalLM architecture class, according to the model card DeepSeek published on Hugging Face. The headline engineering story is the attention mechanism: DeepSeek combined Compressed Sparse Attention [...] DeepSeek V4 Pro 0813 is available two ways: as hosted API access through DeepSeek’s own platform using OpenAI- or Anthropic-compatible request formats, or as a direct download from Hugging Face under the model ID `deepseek-ai/DeepSeek-V4-Pro-0813` for teams with the GPU capacity to self-host. Several router services, including OpenRouter, also proxy the hosted API for teams that want a single integration point across multiple model providers. [...] cache-hit price is close to free. Modality-wise, DeepSeek’s own spec pages describe V4 Pro 0813 as text-only, though at least one third-party comparison mentions experimental image reasoning in some deployments; treat vision support as unconfirmed until DeepSeek documents it directly. Weights are open under the MIT license and hosted on Hugging Face at `deepseek-ai/DeepSeek-V4-Pro-0813`, making it fully self-hostable for commercial use.
Puntos clave: Alibaba introduced Qwen3.8-27B, a lightweight artificial intelligence model optimized to run on personal computers and local hardware. The Chinese tech giant publicly released the weights for its top-tier Qwen3.8 Max model, consolidating its lead in developer adoption over Western rivals. The double launch directly answers Meta’s recent release of its Muse Glimmer model family as both tech giants vie for control of the open-source developer ecosystem. [...] The e-commerce and cloud computing firm unveiled Qwen3.8-27B, saying the compact software delivers strong performance across coding, research, professional tasks, and complex agentic workflows. According to the company, the localized model matches systems ten times its size while running locally on personal devices rather than relying on distant data centers.
Illustration of DeepSeek AI on March 7, 2025. DeepSeek AI is a Chinese artificial intelligence company and the name of its conversational agent, which exploits a large language model. In early 2025, DeepSeek unveils a new model published in open source under the MIT license. Artificial intelligence. (Photo by Riccardo Milani / Hans Lucas / AFP via Getty Images) Hans Lucas/AFP via Getty Images [...] DeepSeek, once known for nearly free AI, has launched its V4 Pro model at prices up to 14 times higher than its budget offerings, charging $1.32/$3.96 per million tokens. This premium pricing, despite V4 Pro's marginal performance improvement over its cheaper V4 Flash, contradicts DeepSeek's prior narrative that AI was becoming too inexpensive. The company, which once caused a massive market dip for Nvidia by suggesting AI infrastructure was unnecessary, is now raising billions for data centers [...] DeepSeek spent the last eighteen months as the market's favorite proof that AI was getting too cheap to justify the money being spent on it. Last Friday the company released V4 Pro, its new flagship, at $1.32 per million input tokens and $3.96 per million output tokens. Its budget V4 Flash model costs $0.14 and $0.28. The company that made its name on nearly free intelligence is now charging up to fourteen times its own floor price for its best model.
DeepSeek V4 Pro 0813, by contrast, is a single open-weight model, a refresh of the earlier DeepSeek V4 Pro release, dated to its August 13 build number. It is DeepSeek’s flagship reasoning and coding model, sitting above the cheaper V4 Flash tier the same way GPT-5.6 Sol sits above Luna. Where GPT-5.6 is API-only and fully closed, DeepSeek V4 Pro 0813 ships weights that developers can download, self-host, or run through third-party inference providers like OpenRouter, which changes the cost [...] DeepSeek’s strongest published reasoning gains, by contrast, cluster around legal and formal-proof tasks, plus general knowledge breadth. The Vals Index puts DeepSeek V4 Pro 0813’s overall score at 52.37%, a jump of 9.48 points over the prior DeepSeek V4 baseline’s 42.89%, moving it to #18 overall on that leaderboard. On the narrower BenchLM DeepSeek-only leaderboard, updated August 19, 2026, V4 Pro 0813 is ranked the top DeepSeek model with a score of 61.3, ahead of every earlier DeepSeek [...] Both models land in roughly the same context window class. GPT-5.6’s three tiers all carry approximately 1.05 million tokens of context, according to OpenAI’s model documentation pages. DeepSeek V4 Pro 0813’s context window is listed at 1.0 million tokens in the most recent cost calculator explicitly tied to the 0813 build, though it is worth noting that some older DeepSeek pricing guides still reference a smaller 256K window for a prior V4 Pro release, a discrepancy that likely reflects the
It was a 27-billion-parameter model from Alibaba: Qwen3.8-27B landed on Hugging Face on Friday under an enterprise-friendly, open source Apache 2.0 license, giving developers downloadable weights for a dense multimodal model. [...] In Alibaba's published comparison table, the 27B model even beats the listed Claude Opus 4.6 Max result on SWE-bench Pro and LiveCodeBench, although Opus remains ahead on Terminal-Bench, GPQA Diamond and Humanity’s Last Exam. Some of Alibaba's evaluations are internal, and benchmark harnesses are not identical across every comparison, making the numbers poor grounds for declaring a universal winner. [...] Alibaba’s strategy of publishing Qwen models across multiple practical size classes has helped make the family a recurring part of developers’ local deployment workflows. Qwen3.8-27B pushes that logic further. Its benchmark scores still need more independent validation, its default reasoning behavior can be painfully inefficient, and no single leaderboard establishes frontier-model parity.
model. A sort of collection of software that determines whether an AI agent can actually get a job done. There are some hard numbers behind the release, too. Harness arrived alongside DeepSeek V4-Pro-0813, the latest version of the company’s flagship model, with much of the attention this time going to its agent capabilities. DeepSeek puts V4-Pro-0813 at 87.9 on Terminal Bench 2.1. It scored 74.1 on Toolathlon-Verified, 71.1 on DSBench-FullStack, and 67.2 on DSBench-Hard. These are DeepSeek’s [...] model. A sort of collection of software that determines whether an AI agent can actually get a job done. There are some hard numbers behind the release, too. Harness arrived alongside DeepSeek V4-Pro-0813, the latest version of the company’s flagship model, with much of the attention this time going to its agent capabilities. DeepSeek puts V4-Pro-0813 at 87.9 on Terminal Bench 2.1. It scored 74.1 on Toolathlon-Verified, 71.1 on DSBench-FullStack, and 67.2 on DSBench-Hard. These are DeepSeek’s [...] There are some hard numbers behind the release, too. Harness arrived alongside DeepSeek V4-Pro-0813, the latest version of the company’s flagship model, with much of the attention this time going to its agent capabilities. DeepSeek puts V4-Pro-0813 at 87.9 on Terminal Bench 2.1. It scored 74.1 on Toolathlon-Verified, 71.1 on DSBench-FullStack, and 67.2 on DSBench-Hard.
## Key Points SpaceX's purchase of Cursor marks its arrival as a serious contender in the software development arena. Grok 4.6 is the first xAI model to rank alongside the best from OpenAI and Anthropic on key benchmarks. SpaceX's cost advantage comes from owning its computing power rather than renting it. [...] Grok 4.6 scored 61 on the benchmark Artificial Analysis Intelligence Index, matching GPT-5.6 Sol and sitting just behind the newest Claude models. On the SWE-bench that ranks large language models as tools for coding, it's in the top cluster. [...] Accessibility Menu ▲ S&P 500 +---%|▲ Stock Advisor +---%Join The Motley Fool AccessibilityHelp The Motley FoolThe Motley Fool SpaceX (SPCX -4.05%) closed its $60 billion acquisition of Cursor last week, just days after releasing its latest AI model, Grok 4.6. The deal gives SpaceX an instant foothold in the enterprise AI market through Cursor's popular code editor.
GuruFocus GuruFocus # Amazon Adds Grok 4.6 to Bedrock Amazon.com Inc. (AMZN, Financials), the cloud and e-commerce giant, is adding SpaceXAI's latest Grok model to Bedrock as competition for enterprise AI customers continues to heat up. Grok 4.6 is now generally available to developers in supported AWS regions, SpaceXAI said. This opens up a much wider path into companies that are already building AI applications on Amazon Web Services. [...] Amazon Bedrock enables customers to access and deploy third-party AI models without having to build the underlying infrastructure themselves. Adding Grok provides AWS users with another option, on top of the growing list of models already available from the platform. But for SpaceXAI, the bigger win is dissemination. AWS could be an opportunity to get Grok in front of enterprise developers and turn model improvements into real commercial usage, not just consumer attention.
## Key Points SpaceX's purchase of Cursor marks its arrival as a serious contender in the software development arena. Grok 4.6 is the first xAI model to rank alongside the best from OpenAI and Anthropic on key benchmarks. SpaceX's cost advantage comes from owning its computing power rather than renting it. 10 stocks we like better than Space Exploration Technologies › [...] SpaceX(NASDAQ: SPCX) closed its $60 billion acquisition of Cursor last week, just days after releasing its latest AI model, Grok 4.6. The deal gives SpaceX an instant foothold in the enterprise AI market through Cursor's popular code editor. [...] An image of Elon Musk in the White House. SpaceX CEO Elon Musk. Image source: The Motley Fool. ## Grok is in the game Grok 4.6 scored 61 on the benchmark Artificial Analysis Intelligence Index, matching GPT-5.6 Sol and sitting just behind the newest Claude models. On the SWE-bench that ranks large language models as tools for coding, it's in the top cluster.
While DeepSeek-V4 Flash achieved high scores in agent-based benchmarks, its success rate was lower on more challenging tasks. Taryn Prambu, a writer specializing in AI and cybersecurity, points out, 'This gap demonstrates why, in enterprise environments, the success or failure of a model depends not on the raw capabilities of the model itself, but on the 'orchestration' that brings multiple systems and tasks together. Even with the same model, the results can vary greatly depending on the [...] Technology analyst Karmi Levy stated that even after the price increase, DeepSeek remains 'far cheaper in every aspect' compared to competing models such as OpenAI, Anthropic, Google, and xAI, and emphasized the importance of appropriately allocating agent AI capabilities. For example, tasks like batch processing are often routine and repetitive, rather than requiring advanced technology, making it appropriate to use a cheaper and more efficient model. Levy said, 'Companies can consider [...] Sanchit Vil Gogia of Greyhound Research, a global technology research and consulting firm, said, 'What's important is whether the AI model's performance is sufficient to meet the actual workflows that companies operate, not whether it's the best at every benchmark. Companies need to determine which combination of model, harness, and provider will get the job done most securely and at the lowest cost.'
xAI’s Grok 4.6 continues a faster release cadence than either rival, with Grok 4.5 having entered beta only months earlier at SpaceX and Tesla before Grok 4.6’s general release. That speed comes with a tradeoff: xAI’s benchmark documentation for Grok 4.6 is the least consistent of the three across independent trackers, and its context window still trails the roughly one-million-token standard that OpenAI and Google have both settled on for their current flagship-adjacent tiers. [...] xAI API curl \ -H "Authorization: Bearer $XAI_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "grok-4.6", "messages": [{"role": "user", "content": "Summarize this contract."}] }'
For example, click the dialog box, select any folder on your local computer, which can be a newly created one or one containing specific project files; after selection, we can put forward requirements to DeepSeek Harness. Generally speaking, you can select the DeepSeek-V4-Pro model and set the reasoning level to Max. Although it costs a little more, it can also reduce the time we spend negotiating with AI.
DeepSeek positions V4 Pro as its agentic and coding-first model, aimed squarely at complex reasoning chains, tool-calling workflows, and long-document analysis. The cheaper sibling, DeepSeek V4 Flash (`deepseek-v4-flash`, updated to the 0731 build on July 31, 2026), trades some reasoning depth for a 5x higher concurrency ceiling and a fraction of the per-token cost. We already covered Flash setup in a dedicated DeepSeek V4-Flash tutorial; this guide focuses on the Pro tier and the extra steps [...] DeepSeek V4 Pro’s positioning as an agentic model means tool calling (also called function calling) is where it’s meant to earn its higher price over Flash. The API follows the same tool-calling schema as OpenAI’s Chat Completions format, so if you’ve built agents against GPT models before, the pattern will look familiar. Here’s a minimal agent that gives the model access to a weather-lookup function and lets it decide when to call it:
GPT‑5.6 is our strongest model yet for accelerating AI research. Inside OpenAI, researchers use it across the development loop: diagnosing failures, optimizing training systems, running experiments, and interpreting results. We already saw that acceleration and stronger adoption during the internal testing period of GPT‑5.6, as average daily output tokens per active researcher were more than twice the highest level observed for GPT‑5.5.
.5.5 6 5..6 6 5 5.5 5 5 6 5 6.6 6.5 6...6..5 6 5 5.5 6.6 6 5 6 [...] 5 .5 5 5 5 5 5 5 55 5 6 5. 5 6 5 55 5 5 5 6 5 6 5 5 . 5 66 5 . 6 5 . 5 . 5 6 5 5 65 [...] 5 6 5 6 6 6 5 5 5 5 5 5 5 5 6 5..5.5 5 5 5 5 5 5..5 5..5 5 5 5 5 6 5 6 5 5 5 5.6 5 5.6 5 5 5 5 6 5 5.5 6 5 5.5 5 5 5 5 6 5 5 6 5 6.5 5 6 5 5 5 5.5 5 5 5 5.5 5 5 5 5 5 6 6 6 5 5.5 5 5 5 5.6 6 5 5 5 5 6 6 5 5 5.5 5 6 6 5 5 5 6 5 5 5 6 5 5 5 6 5 5 6 5 5 5.5 5.6 6 6 5 5 5 5 5 5 6 6 5.5 5 5 5
# Indian IT Sector Pivots To Performance Billing As AI Demands Price Cuts. India's IT services sector is moving from hourly billing to performance-based contracts as AI reduces the need for human labor. Clients are demanding price cuts of up to 30%, which is putting pressure on profit margins. The Indian IT services industry is undergoing a structural change as artificial intelligence (AI) forces a departure from the traditional model of charging clients by the hour. They are instead demanding contracts tied to specific business outcomes, effectively forcing IT firms to do more work for less cost. Clients are increasingly asking for price reductions in the range of 25% to 30%. This means the IT firm's revenue is now linked to the value it provides, such as efficiency gains, rather than the number of hours its employees work. IT firms are facing revenue headwinds as legacy managed-services contracts shrink faster than new, AI-led projects can scale.
Google on Wednesday announced a slew of new study tools across Search and Gemini, including AI-generated interactive visuals, 3D simulations, a decided student hub, customized practice quizzes, and more. The launch of the new study features marks Google’s latest effort to make Gemini the AI assistant that students turn to when learning and studying, as it competes with companies like OpenAI and education startups such as Knowt and Gauth, which are also offering their own learning and practice tools. On Search, students can now generate custom tools and simulations to help them understand complex topics. For example, if a student is learning about pH levels, they can search for “pH scale” and get an interactive visual in an AI Overview that makes the basics easier to understand. To go even further, they can ask a follow-up question for something more specialized, like plotting citrus fruits on the pH scale, and AI Mode in Search will then create an interactive experience tailored to the question.
Corporate expense management platform Ramp is hot on the heels of Stripe in setting up toll houses for AI inference. Ramp on Wednesday evening launched its own AI model routing service, dubbed Router, that lets users and companies use and switch between various large language models through an API. It’s free to use for the remainder of 2026 (users will still have to pay for AI model inference costs), and it comes with a $26 credit launch offer. In its function, Router is pretty similar to how OpenRouter operates, though the latter offers many more AI model options than Ramp’s current offerings. For Ramp, entering the model routing business offers a two-pronged opportunity: it gets to tap the booming AI inference market, and *also* offer its existing clients a model routing service that fits in neatly with its existing products, which includes AI token usage monitoring and token spend management.
techcrunch.com· 22h agoGoogle said to end China Pixel production in 2027 - Breaking The News. # Google said to end China Pixel production in 2027. Google plans to move all manufacturing for its Pixel devices, including smartphones, watches, and wireless earbuds, completely out of China starting next year amid ongoing tensions between Washington and Beijing, Nikkei Asia reported, citing sources familiar with the matter. To achieve this, the tech company has been expanding its production capacity in Vietnam and India. While a significant portion of Google's hardware is still made in China, this transition marks a major strategic shift in its global supply chain operations. Alphabet, Inc.'s Google announced on Thursday that its launching Gemini 3.7 Flash, calling it its "most intelligent... ##### Google: Pixel 11 display brightest of all smartphones. Alphabet Inc.'s Google insisted during its Made by Google 2026 event that the display of its newest smartphone... Google unveiled the Pixel Watch 5 on Wednesday, featuring a smarter Gemini experience, advanced health... Google LLC officially launched its 11th generation of Pixel smartphones, including the Pixel 11, Pixel 11 Pro,...
A new dedicated study hub includes a study notebook, customized flash cards and practice quizzes. It’s back-to-school season, and Google is getting in on the academic action. The company announced on Wednesday that a dedicated student hub, featuring a suite of AI study tools, will now be available on its Gemini AI assistant and through Google Search. The dedicated study hub can be found within Gemini and contains a suite of tools, including a study notebook, customized flash cards and practice quizzes to help with research and assignments. Students will be able to ask Gemini to dig into a topic they’re researching and then switch tasks while the tool compiles the information in the background. When the report is finished, Gemini will send a notification and be available to discuss the report’s details and participate in a question-and-answer session to help the student better understand the findings. Stick with AI Mode in Google Search, and you can ask follow-up questions about the topic in question to dig deeper into a subject.
Large language models (LLMs), including the models behind ChatGPT and Claude as well as numerous other systems developed specifically for medical applications, are increasingly used in clinical workflows. However, their adoption is outpacing the development of systems for oversight and safety. An interdisciplinary team of researchers at the Else Kröner Fresenius Center (EKFZ) for Digital Health at TUD Dresden University of Technology and University Hospital Dresden, together with national and international colleagues, has systematically analyzed the risks associated with LLM use in medicine. The review, published in Nature, brings together evidence from medical AI, cybersecurity, regulatory science, ethics and behavioral psychology and outlines strategies for trustworthy and responsible use of artificial intelligence (AI) in clinical practice. LLMs have the potential to support and enhance the work of healthcare professionals in a variety of areas. The authors of the newly published review show that these risks can arise throughout the entire lifecycle of AI systems: from initial model design to training data, model deployment, and real-world use in clinical environments.
miragenews.com· 2d ago# Gemini Live adds Deep Research as Notebooks come to AI Mode. With the back-to-school season underway, Google is introducing new study tools across the Gemini app and Search. College students in the US can get one free year of Google AI Pro ($19.99 per month), while there’s a bundle with YouTube Premium that offers up to 70% off. A new “Student” hub (gemini.google.com/students) in the Gemini side panel surfaces offers and tools, like flashcards and practice quizzes. Meanwhile, Google AI Overviews and AI Mode are adding a host of study tools. * “To go even further, you can follow up and ask to see something more specialized, like plotting citrus fruits on the pH scale, and AI Mode will create a customized experience just for your question.”. AI Mode is getting integration with Gemini Notebooks. Google Lens is getting a new interactive learning experience that can discuss “tough concepts and confirm if you’re on the right track.” Available in the coming weeks (globally in English), you just snap a photo of the problem.
