Arena analyzed 26,225 Agent sessions, 30 agents, and 13 labs from September 12 to October 23, 2026, and found that if billed at the full input price, the typical task cost for the top ten agents would be 3.6 times higher, and 592% higher on MiMo V2.6 Pro. For agents, 86% to 96% of tokens are repeatedly read context, while their own output accounts for only 1% to 2%. Cache invalidation increases the typical task cost by about 5%, and on Claude Fable 5.1, 1 in every 10 sessions sees an increase of more than 18%.
Arena analyzed 26,225 Agent sessions, 30 agents, and 13 labs from September 12 to October 23, 2026, and found that if billed at the full input price, the typical task cost for the top ten agents would be 3.6 times higher, and 592% higher on MiMo V2.6 Pro. For agents, 86% to 96% of tokens are repeatedly read context, while their own output accounts for only 1% to 2%. Cache invalidation increases the typical task cost by about 5%, and on Claude Fable 5.1, 1 in every 10 sessions sees an increase of
TELUS Digital built an outbound voice agent based on ElevenAgents that proactively calls customers who have just activated home internet, resulting in a 30-day churn rate for customers who received the call that is less than half that of new customers overall. The agent can disclose its AI identity, explain the first bill, guide customers to self-service tools, and remind them of bundle discounts, and can authenticate customers during the call via SMS verification code and read real-time billing data through an API. TELUS Digital selected ElevenAgents after evaluating options from November 2025 to February 2026 against five criteria: voice expressiveness, agent framework flexibility, real-time emotional adjustment, enterprise security and compliance, and response latency.
TELUS Digital built an outbound voice agent based on ElevenAgents that proactively calls customers who have just activated home internet, resulting in a 30-day churn rate for customers who received the call that is less than half that of new customers overall. The agent can disclose its AI identity, explain the first bill, guide customers to self-service tools, and remind them of bundle discounts, and can authenticate customers during the call via SMS verification code and read real-time billing
Zhaopin launched its self-developed B2B recruiting Agent "Zhaopin Xiaozhao" at the 2026 Tencent Global Digital Ecosystem Summit, and partnered with office collaboration platform WorkBuddy to integrate it into its workspace for the first time, offering an initial open experience to enterprise users. The Agent can proactively take over the process from sourcing candidates, organizing resumes, and assisting evaluation to recruiting communication and scheduled task advancement, achieving "more accurate with use" based on data from 15.49 million+ cumulative partner enterprises and 404 million+ professionals. HR can generate job JDs with one click in WorkBuddy and publish them to the Zhaopin platform, with "Zhaopin Xiaozhao" continuing the subsequent sourcing, screening, and communication services.
Zhaopin launched its self-developed B2B recruiting Agent "Zhaopin Xiaozhao" at the 2026 Tencent Global Digital Ecosystem Summit, and partnered with office collaboration platform WorkBuddy to integrate it into its workspace for the first time, offering an initial open experience to enterprise users. The Agent can proactively take over the process from sourcing candidates, organizing resumes, and assisting evaluation to recruiting communication and scheduled task advancement, achieving "more accur
Tripo introduces step 2 of the Tripo Agentic Canvas automated 3D workflow guide: before execution, the Agent first confirms a step-by-step plan with the user, who can review, modify, or directly approve it, and can switch between Control / Auto modes to decide how much freedom to give the Agent.
Tripo introduces step 4 of the Tripo Agentic Canvas workflow: models can be exported as GLB, USD, FBX, OBJ, STL, or 3MF, or sent directly to tools such as Blender, Maya, Unity, and Unreal via DCC Bridge; after sharing a project link, others can clone the entire workflow.
Ayì paraphrases Stripe founder John Collison's 52-minute interview with OpenAI CEO Sam Altman, saying Altman mentioned that an internal new model had cracked frontier math problems, with the proof related to the Navier-Stokes equations consuming on the order of $10 million in compute. Altman also said he prefers Agent Teams rather than swarms to describe multi-agent collaboration, and believes sandboxes can hardly contain superintelligence, so one can only rely on internal monitoring and alignment. He also criticized Silicon Valley's current AI narrative as bad salesmanship, insisting on iterative deployment to avoid a small elite deciding the future for the whole world.
l new model had cracked frontier math problems, with the proof related to the Navier-Stokes equations consuming on the order of $10 million in compute. Altman also said he prefers Agent Teams rather than swarms to describe multi-agent collaboration, and believes sandboxes can hardly contain superintelligence, so one can only rely on internal monitoring and alignment. He also criticized Silicon Valley's current AI narrative as bad salesmanship, insisting on iterative deployment to avoid a small e
Serro has released "Agentic TPM," which keeps coding agents in sync when the decisions they rely on change. It connects to Slack, Linear, GitHub, Jira, Zoom, Notion, email, and code repositories, maintains a versioned record of each project's decisions, owners, risks, and history, and treats agent failures as coordination problems.
indigo says they've recently grown very fond of the Markdown editor Dawn, calling it "built for Agents." It supports md, cvs, json and jsonl, and lets you preview and edit these files without generating html, making it suitable for managing local Agent directories and knowledge bases. It supports Mac and iPhone.
Alibaba Cloud has open-sourced the "AI Agent Handbook," covering five parts—architecture, building, running, governance, and tuning—with Harness at its core to move enterprise Agents from pilot to governable production. The 2026 survey cited in the handbook shows that about 46% of enterprises have completed or are in the process of development, while only about 18% have actually deployed to production; about 90% have context and memory needs, about 63% have multi-model routing and cost governance needs, and fewer than about 8% have trajectory-based automated evaluation. Those paying attention to or implementing MCP total about 37%, and only about 9% have private deployments within the enterprise.
Alibaba Cloud has open-sourced the "AI Agent Handbook," covering five parts—architecture, building, running, governance, and tuning—with Harness at its core to move enterprise Agents from pilot to governable production. The 2026 survey cited in the handbook shows that about 46% of enterprises have completed or are in the process of development, while only about 18% have actually deployed to production; about 90% have context and memory needs, about 63% have multi-model routing and cost governanc
Vercel CEO Guillermo Rauch published a post explaining how Vercel CI solves the AI coding bottleneck, in three layers: fast and stable infrastructure (20 clusters worldwide, running millions of microVMs per day), Turborepo integration (automatically computing the dependency graph and skipping and caching existing work), and Agentic CI. He said that vercel ci in the CLI is the first step; agents can now inspect, abort, trigger, and replay CI, and Vercel Agent can assist with optimization. More capabilities will be added later, giving CI engineer-like judgment. The official Vercel account previously announced that Vercel CI had entered beta, with CI in internal repositories sped up 2-6×, and optimized for agents via Vercel MCP and the CLI.
ure (20 clusters worldwide, running millions of microVMs per day), Turborepo integration (automatically computing the dependency graph and skipping and caching existing work), and Agentic CI. He said that vercel ci in the CLI is the first step; agents can now inspect, abort, trigger, and replay CI, and Vercel Agent can assist with optimization. More capabilities will be added later, giving CI engineer-like judgment. The official Vercel account previously announced that Vercel CI had entered beta
Gemini 4 Carbon (High) took first place in Text Arena with 1528 points, surpassing Gemini 4 Argon (High)'s 1520 points. It ranks second in Agent Arena, behind only Claude Opus 5.5, and has entered a new Pareto frontier; in Web Dev Arena it ranks fourth with 1768 points, two points behind third-place Claude Sonnet 5.5. Arena data states its median cost per Agent Arena task is $1.80, 72% lower than top-ranked Claude Opus 5.5 (Max). Its Artificial Analysis V5 score is 59.
Snaptrude has launched a building design Agent that can start from a site or a brief to generate LOD 400 construction-level models including structure, MEP and complete drawings, and supports exporting back to existing software while preserving component families. During early access, more than 2,000 designers have run over 25,000 tasks, and the company has also released the AECBench benchmark along with its first batch of results. Rohan Paul relayed that most architectural AI stops at the level of form, whereas this model carries actual components that can be directly manufactured and installed.
Snaptrude has launched a building design Agent that can start from a site or a brief to generate LOD 400 construction-level models including structure, MEP and complete drawings, and supports exporting back to existing software while preserving component families. During early access, more than 2,000 designers have run over 25,000 tasks, and the company has also released the AECBench benchmark along with its first batch of results. Rohan Paul relayed that most architectural AI stops at the level
DeepLearning.AI has launched a new Context Engineering course, taught by Chroma CEO Jeff Huber and CMPND CEO Drew Breunig, framing context engineering as the main thread for keeping Agents reliable and token-efficient in complex multi-step tasks. The course builds a harness from scratch in Python and uses traces and evals to quantify each change, covering three capabilities—Agentic Search, Context Management, and Agentic Memory—across four modules, with labs mostly consisting of about 1-hour coding exercises. The course assumes intermediate Python, familiarity with LLM applications and tool calling, and basic retrieval.
epLearning.AI has launched a new Context Engineering course, taught by Chroma CEO Jeff Huber and CMPND CEO Drew Breunig, framing context engineering as the main thread for keeping Agents reliable and token-efficient in complex multi-step tasks. The course builds a harness from scratch in Python and uses traces and evals to quantify each change, covering three capabilities—Agentic Search, Context Management, and Agentic Memory—across four modules, with labs mostly consisting of about 1-hour codin
AgentWeb launches, enabling AI agents to complete web tasks without a browser by invoking preset actions instead of reading screenshots and clicking buttons. The company claims it completed 47/47 real-time tasks across 15 government portals, 7 insurance companies, and 25 login-required sites including Amazon and LinkedIn, with a median time of 1.9s and a total cost of $3 for the full round. Each mapped task is exposed as a tool, including an input schema, preconditions, permission boundaries, and completion proofs, and can be called from any assistant.
AgentWeb launches, enabling AI agents to complete web tasks without a browser by invoking preset actions instead of reading screenshots and clicking buttons. The company claims it completed 47/47 real-time tasks across 15 government portals, 7 insurance companies, and 25 login-required sites including Amazon and LinkedIn, with a median time of 1.9s and a total cost of $3 for the full round. Each mapped task is exposed as a tool, including an input schema, preconditions, permission boundaries, an
indigo published a post teasing this installment of its month-end livestream, themed "Evolving Toward the Agentic Internet," saying it had originally planned to keep things simpler but the content became super rich, and inviting readers to check out the agenda and reserve a spot; it will go live on YouTube over the weekend.
Tencent, Tencent Cloud, and WorkBuddy have jointly released the 78-page "2026 AI Agent Enterprise Solutions and Implementation Practice Handbook," which uses WorkBuddy as the entry point to advance personal assistants into "digital employees" that enter business systems, with ClawPro, Agent Runtime, and TokenHub responsible for provisioning, isolation, models, and governance. The handbook proposes a three-stage framework of "able to converse → able to work → able to take a post," and presents case data including Yozoo's Agent penetration rate exceeding 75%, Youka's MTTR of 5–15 minutes, and operations and maintenance costs of approximately -40% (all as self-reported in the text).
Tencent, Tencent Cloud, and WorkBuddy have jointly released the 78-page "2026 AI Agent Enterprise Solutions and Implementation Practice Handbook," which uses WorkBuddy as the entry point to advance personal assistants into "digital employees" that enter business systems, with ClawPro, Agent Runtime, and TokenHub responsible for provisioning, isolation, models, and governance. The handbook proposes a three-stage framework of "able to converse → able to work → able to take a post," and presents ca
Shao Meng believes that although domestic Video Agent teams such as LibTV, TapNow, Flova, Pexo, and Medeo emphasize barriers like ecosystem, Agent architecture, and industry know-how in financing roadshows, after all their work they may have all ended up as token relay stations. He speculates that if these websites supported BYOK, the cost for users to bring their own video generation model API Key would be lower than consuming Credits on the site, and even lower when the API Key has a discount. He suggests using REA or building your own Video Agent from scratch paired with discounted API Keys to reduce costs; for users creating video content on TikTok, Ins, and Douyin, the money saved would be considerable.
Shao Meng believes that although domestic Video Agent teams such as LibTV, TapNow, Flova, Pexo, and Medeo emphasize barriers like ecosystem, Agent architecture, and industry know-how in financing roadshows, after all their work they may have all ended up as token relay stations. He speculates that if these websites supported BYOK, the cost for users to bring their own video generation model API Key would be lower than consuming Credits on the site, and even lower when the API Key has a discount.
The MiniMax Code team has open-sourced MCode Stack, writing the Agent workflows and Skills the team uses daily into installable SKILL.md files and manifests, which can be integrated as plugins into MiniMax Code and any Agents that support Skills. The repository packages stacks by member, among which stacks/ronny/ is the most complete, containing 15 skills covering sub-Agent decomposition, context and tool budget auditing, long-session handoff, independent review, and hotfix judgment. The companion notes/ column MCode Engineering Notes explains the rationale; the six engineering-* entries are adapted from Google Engineering Practices, with the adapted portions under CC-BY 3.0 and the rest under MIT.
The MiniMax Code team has open-sourced MCode Stack, writing the Agent workflows and Skills the team uses daily into installable SKILL.md files and manifests, which can be integrated as plugins into MiniMax Code and any Agents that support Skills. The repository packages stacks by member, among which stacks/ronny/ is the most complete, containing 15 skills covering sub-Agent decomposition, context and tool budget auditing, long-session handoff, independent review, and hotfix judgment. The compani
A 2026 OpenAI SWE interview retrospective shows that the process includes a recruiter screen, two technical screens of about 60 minutes each on the same day, a take-home assignment of about 48 hours, a technical deep dive that probes the code line by line, and 4 onsite rounds. The onsite coding includes a progressive multi-part problem (such as token-level streaming diff and rollback), and the system design question is "Design ChatGPT," involving GPU allocation and autoscaling under non-stationary traffic. There is also an optional fifth round, Agentic Coding, in the beta stage, which requires using an AI coding agent to handle problems in large codebases that are difficult to address by hand.
technical deep dive that probes the code line by line, and 4 onsite rounds. The onsite coding includes a progressive multi-part problem (such as token-level streaming diff and rollback), and the system design question is "Design ChatGPT," involving GPU allocation and autoscaling under non-stationary traffic. There is also an optional fifth round, Agentic Coding, in the beta stage, which requires using an AI coding agent to handle problems in large codebases that are difficult to address by hand.
