Results for “agents”

…given for llama-server / Unsloth docs) (danielhanchen). Coding agents + reliability + “availability is the new frontier”Codex 5.3 and coding eval chatter: Anecdotal reports of Codex 5.3 solving “promising” tasks and pushing benchmarks like …

…can opt in/out of email frequencies!AI Twitter RecapAI Coding Agents, Developer Tooling, and the Race to Own the IDECursor’s Composer 2 looks like the day’s biggest developer-model launch: @cursor_ai released Composer 2, positioning it as a…

…Claude Fable 5 at 60 and ahead of Opus 4.8 at 56. On coding agents, AA later reported K3 scoring 57 on its Coding Agent Index, matching GPT-5.6 Terra and GPT-5.5, ahead of Opus 4.8, with 84% Terminal-Bench v2, 64% DeepSWE, and 23%…

…push toward composite systems rather than monolithic one-shot agents. Google’s Gemini 3.5 Flash Cyber as a graph-engineering case study: One of the more substantive takes on Google’s cyber release came from @Kseniase_, who highlighted…

…an on-device complement to—not a replacement for—frontier cloud agents (launch, specifications). Claimed results: Recursive reports 54.9% on Terminal-Bench 2.1, 89.7% on SWE-bench Pro v2 and 74.2% on SWE-bench Verified. These are release…

…reminder to verify tool call requests: @jeremyphowardCoding agents in practice (workflow shift, prompt caching, “app store” thesis) Karpathy’s “bespoke software” vignette: Claude reverse-engineers a treadmill API to build a custom…

…for a subset of tasks, the RL reward signal as well. Research agents collect task patterns from real work and turn them into runnable long-horizon environments with multi-step dependencies and hidden state; a judge agent then attempts each …

…AutomationGoogle Open Sources Harness2 for Self Improving AI Agents (Activity: 1839): The image shows Google’s Harness² project page for “Recursive Agent Harnessing for an Open World,” matching the post’s claim that Google open-sourced a…

…and businesses as possible with different superintelligent agents aligned to their goals that check and compete with each other in the ways our natural economy behaves. This balance would be further enhanced if there were multiple frontier …

…leaderboard for Zapier’s AutomationBench, evaluating agents across 657 tasks and 40 simulated SaaS apps with both objectives and guardrails. Claude Fable 5 led at 48.6%, narrowly ahead of Opus 4.8 at 48.5%, with Gemini 3.5 Flash at 42.6%…

…memory, RTX GPU—fits the same trend from the hardware side. Agents, Harnesses, and the Shift from Frameworks to Execution LayersThe center of gravity is moving from “frameworks” to agent harnesses and execution environments: Several posts…

…autoscaling, backups, forks, and control-plane observability. Agents, Harnesses, and System Design Becoming the Real Product LayerHarnesses were a recurring theme across builders: Harrison Chase’s conversation with Factory AI’s Eno Reyes…