AI Coding Live News

When autonomous LLM coding agents tackle complex software maintenance tasks, early minor missteps frequently compound into unrecoverable cascading failures. Existing runtime monitors struggle due to static guidance schemes: fixed-interval periodic checks introduce substantial token overhead during healthy executions while failing to intervene promptly during critical errors, and heuristic rule-based monitors fail to detect subtle reasoning errors in natural language. Researchers from Singapore Management University and University College London introduce Cadence (arXiv:2610.12269), a dynamic runtime monitoring framework delivering strategic guidance conditioned on real-time execution health. Cadence integrates a Two-Tier Intervention Module—supplying non-blocking advisory-level advice for minor lapses while enforcing replacement-level overrides (discarding erroneous actions and mandating step regeneration) during severe breakdowns—with an adaptive Inspection Scheduler that tightens supervision after replacement actions and relaxes monitoring intervals following advisory guidance. Evaluated on 300 SWE-bench Lite tasks across mini-swe-agent and Moatless, Cadence improves task resolve rates by 25.33% (+76 tasks) and 15.67% (+47 tasks) respectively over unmonitored baselines, while achieving up to 29.93% token savings compared with leading runtime monitors.

Contemporary embodied intelligence relies heavily on neural network policies: foundation models either run end-to-end Vision-Language-Action (VLA) architectures or repeatedly query Vision-Language Models (VLMs) at high frequencies via agent harnesses. This paradigm imposes severe inference latency, costly deployment bills, and unpredictable non-deterministic execution failures. Researchers from S-Lab at Nanyang Technological University (NTU Singapore) introduce a paradigm shift (arXiv:2610.12369): Embodied Turing Machines and Code-Only-as-Policy (COAP). Under this formulation, the physical environment operates as a Turing machine where the tape encodes robot proprioception and visual states, while the transition rules are executed entirely by deterministic, stateful code. A shared, modular code library governs decisions across episodes without running any neural network models at test time. Crucially, explicit stateful code provides the optimal medium for Recursive Self-Improvement (RSI): autonomous coding agents iteratively author, debug, and refactor robot skills in closed simulation loops. Across 42 complex bimanual manipulation tasks in RoboDojo, the synthesized COAP library reaches a 70.24% success rate with zero neural models running at runtime, slashing deployment latency and cost to zero.

Tool-using agents execute consequential modifications to external systems, yet nominal task success frequently conceals unverified actions executed without prior evidence. Researchers from the National University of Singapore (NUS) present a rigorous empirical study on the breakdowns along the evidence-to-action chain (arXiv:2610.07753). Evaluating ten model-harness configurations reveals a striking paradox: high static action assessment accuracy routinely masks brittle interactive execution. Failures overwhelmingly originate prior to execution: agents terminate prematurely before collecting requisite evidence, or initiate state-altering actions before foundational proofs are established. The authors formulate SafeActBench, spanning 656 rigorous cases across six operational domains and five protocols progressing from static judgment to complex multi-action workflows. Supported by a provenance-bound Evidence Ledger and deterministic trajectory evaluator, findings demonstrate that while single-step execution is stable once evidence is verified, multi-action workflows frequently collapse under unresolved prerequisites and truncated state transitions.

Anthropic announced the opening of its APAC headquarters in Singapore, appointing Dale Finlay as General Manager for ASEAN. The hub will serve financial institutions, governments, and engineering teams, scaling enterprise deployments of Claude Fable and Mythos.

Ollama’s Pro, Max, and Team plans now bill at published per-token rates against a monthly credit pool: $20/month with $60 of usage, $100 with $300, and $500 Team with $1,000 shared across unlimited users. The plans plug into Claude Code, Codex, and a first-party API, with no service fees and no 5-hour or weekly caps—when the pool runs out, traffic continues at the same token price. Inference stays on dedicated US/EU compute (Singapore for a limited Qwen set) with zero retention and no training on prompts. Existing subscribers can keep the old plan; new signups land on the token model because GPU-time billing became hard to forecast as open models scaled to sizes such as Kimi K3.

DeepMind ran Gemini Flash Lite in a confidential setup so evaluators cannot see weights and Google cannot see prompts. Singapore’s AI Safety Institute is among the partners. No scores published.