로컬 실행, 에이전트 운영, 저비용 추론으로 이동하는 AI 스택

Great to see Perplexity bring Portable Computer to Windows. As more AI experiences move closer to the device, we're excited to see developers build new experiences that take advantage of local AI compute.

dar

@radbackwards

1일 전

Apple should be hyper obsessed with creating great private/local AI— that should be the only goal.

Hermes Agent is open for business. Nous Portal now lets you invite colleagues to a Hermes Business account: your team gets agents across channels while sharing one central balance with per-member caps and shared skills that compound into proprietary IP. Hermes Enterprise brings the same capabilities to on-prem or the cloud of your choice: a complete, self-improving, sovereign AI stack already trusted by some of the world's largest companies. Contact us to join them. https:// portal.nousresearch.com/business?utm_s ource=twitter&utm_medium=social&utm_campaign=business_launch&utm_content=tweet_2026-09-08 …

@googlecloud and Inferact are announcing today a partnership to make TPU a first-class citizen in @vllm_project . This partnership puts both teams on one engineering roadmap to bring TPU to the broader open model ecosystem, optimizing vLLM as the agentic production serving engine for TPU: • Production serving features and optimized kernels • A native PyTorch path via TorchTPU • Moving towards day-0 support for frontier model releases We're also launching a community program: shared TPU capacity for open-source contributors, plus dedicated review and design help from the core vLLM maintainers at Inferact. Everything this collaboration produces is open source. Read the full announcement: https:// inferact.ai/news/google-tp u-partnership …

Elon Musk

@elonmusk

1일 전

And assembly with direct calls to hardware accelerators in the NVidia SMs

Viv

@Vtrivedy10

1일 전

Alex & the Tenex team rock Evals & Environments & Data are driving a huge chunk of the world’s AI progress from the frontier labs, to open source, to every company serving agents to real users. At @LangChain we want to help every team own the process of turning their valuable data into evals that improve their agents over time - Evals are basically Tasks/Tests that help you measure how an agent does on work you care about - Evals today look different than 3 years ago (no more prompt in/out). Today’s agents do much more autonomous work and we need Environments that faithfully capture the real work they do - Building good evals today is still hard, this is why frontier companies pay millions to acquire them across domains - Every team looking to Own their Intelligence will need to own their Evals. Both so they can measure the progress of agents on their tasks and also because Evals are literally the thing that shapes what your agent will become. - When you train on Evals or hill-climb them with harness-engineering, you transfer the behaviors from the eval into your agent. Garbage in garbage out - the inescapable lesson of machine learning there’s so much to gain from helping every team own their Data to Eval pipeline, let’s cook

Patrick Erichsen

@Pat_Erichsen

1일 전

as you become more ambitious with coding agents, delegation becomes increasingly important the new "suggested task" feature in @openclaw gives your agent a tool to do this for you if the agent identifies a sufficiently well-scoped piece of work, it will recommend kicking off a new session to work on it

DeepSeek-V4.1-Flash (Max) is a breakthrough in performance to cost efficiency. With +4.87% net improvement at $0.07 cost per median task, it’s reshaped the Pareto frontier for Agent Arena! Among the top 3 open models, DeepSeek-V4.1-Flash (Max) has the lowest median task cost. For comparison, it retains: - 98% of Hy4 preview’s net improvement, at 73% lower cost - 76% of Kimi K3 (Max)’s performance, at 92% lower cost. Against models as powerful as Fable 5 or stronger, DeepSeek-V4.1-Flash (Max) retains 35–54% of their net improvement at 97–99% lower cost. Those top models cost 37–76× more per task. Net improvement over Arena baseline | Median cost/task: - Claude Fable 5.1 (Max): +13.90% | $4.54 - GPT 6 Astra (Max): +11.90% | $4.09 - Claude Opus 5 (Max): +11.09% | $3.52 - Claude Opus 5 (High): +10.49% | $2.24 - Claude Fable 5 (High): +9.03% | $2.19 - Claude Opus 4.8 (High): +7.75% | $1.36 - GPT 5.6 Sol (xHigh): +7.40% | $1.09 - Kimi K3 (Max): +6.39% | $0.77 - Hy4 preview: +4.96% | $0.22 - DeepSeek-V4.1-Flash (Max): +4.87% | $0.06 With this release, GPT-5.6 Luna (xHigh), GLM-5.3-Flash, and DeepSeek-V4-Flash fell off the Pareto frontier for Agent Arena. Congrats again to the @deepseek_ai team on this release!

We’re excited to see Dario and Sam commit to “embedded evaluators who have employee-like access to verify safety practices and report incidents.” [...] “whose role is to verify adherence to safety practices and commitments, report incidents, and help assess the alignment of not just completed AI models but training pipelines and processes.” We’ve long advocated for third-party evaluators to have deep access. Risks from internal deployment and scheming cannot be appropriately assessed without direct insight into the training pipelines and internal deployment practices. Apollo is looking forward to working on Embedded Evaluations with all Frontier AI developers. 1/3

A lot of alignment issues would be sorted if companies had to be 100% transparent on datasets used for training models. No cyber or biological (outside standard medical) data allowed for base models Tuned versions with that data shown to and monitored by evaluators