The Groundy archive · Page 28 of 38
Browse Groundy's complete archive of 895 articles on AI, developer tools and infrastructure. Page 28 of 38.
649–672 of 895 articles · Newest first
Vertical finance benchmarks show agents break on chained calculations, and accuracy fails independently of data-handling safety. Lenders must require audit-shaped evidence.
NLnet funds open-source infrastructure through equity-free grants from a 1997 endowment, a model that sidesteps venture capital but depends on EU programme continuity.
Adam's YC-backed Text-to-CAD tool markets itself as open-source AI CAD, but no source confirms a repo or license, and its accuracy on engineering geometry stays unproven.
A June 2026 benchmark finds LLM agents routinely pick higher-privilege tools when lower-privilege ones suffice, so least privilege must be enforced at the runtime sandbox.
Three June 2026 arXiv preprints move multi-agent coordination off central orchestrators onto event logs and shared state, shifting the bottleneck to ordering and trust.
Epic open-sourced Lore, a version control system pitched at the scaling ceiling where Git's object model strains. Operators should demand benchmarks before any migration.
UltraQuant cuts agent time-to-first-token 3.47x with 4-bit KV caching on AMD CDNA4, but its June 2026 preprint omits the accuracy numbers operators need to ship it.
A preprint shows defensive misdirection can cut estimated attacker-success bounds by up to two orders of magnitude, but leaves the cost to legitimate tasks unmeasured.
Two June 2026 preprints find RL training in LLM agents rewards the wrong behavior, widening the safety gap and inflating SWE-bench scores by 14 points.
Anthropic's record splits AI procurement risk in two: model behavior on benign prompts versus vendor refusal. Both block deployments but need different diligence.
TuneAhead forecasts fine-tuning success from a short simulated probe, catching 89.4% of winners and 91.0% of failures on Qwen2.5-7B-Instruct for 58.4% compute savings.
A June 2026 preprint makes the employer profit objective in gig hiring explicit, surfacing how optimized dispatch can shift timing risk onto contingent workers.
LLM-written CUDA kernels compile, run, and pass smoke tests while returning wrong numerics, so crash-free execution is not enough to trust AI-generated GPU code.
The RoboSSM preprint swaps the transformer backbone of in-context robot imitation for a Longhorn state-space model and claims LIBERO gains. The full paper is still pending.
Routing architecture decides whether MoE experts can be pruned safely: hard routers keep calibration per expert, soft routers need an aggregate guard.
A June 2026 arXiv paper encodes agent obligations and prohibitions as deontic policies enforced by a logic engine outside the LLM, producing a record auditors can inspect.
GLM-5.2 and Kimi K2.7 Code shipped a day apart as open-weight coding models. One leads the open-weights index; the other wins on price and a claimed token-efficiency gain.
Per-block linear recoverability in transformer FFNs swings from near-linear to strongly nonlinear between layers, so compression tools must probe each block per checkpoint.
Cursor is being acquired by SpaceX for $60B. Windsurf was absorbed into Cognition and rebranded Devin Desktop on June 2. Both vendors have changed; the tools have too.
A June 2026 preprint finds essay quality is linearly decodable from LLM internals, but cannot show whether that signal tracks argument quality or just length and fluency.
Every GLM-5.2 benchmark figure, from 62.1% SWE-bench Pro to 99.2% AIME, traces to Zhipu's own evaluations. This guide separates vendor-run scores from external evals.
GLM-5.2 ships under MIT, removing the Llama usage-threshold audit burden, but finance and healthcare teams still face compliance gaps when self-hosting this 753B MoE model.
GLM-5.2 scores 81.0 on Terminal-Bench 2.1, trails Claude Opus 4.8 (85.0) on shell tasks, but wins on 1M-token monorepo context. Here is how to route tasks.
GLM-5.2 posts 62.1% on SWE-bench Pro and 81.0 on Terminal-Bench 2.1, four points behind Opus 4.8. MIT weights are self-hostable; flat plan starts at $18/month.