#llm

This article is a copy-edited transcript of the presentation given at DevFest Melbourne, October 3,...

In RLVR the verifier is your loss function. How to build deterministic gyms: 3-layer design, mutation checks, cache isolation, and reward-hacking defenses.

An LLM gateway is a small system with a big trust load. It's small enough for one on-call to own, and...

Free API tiers get farmed. That is not news. What surprised me was how much the farming looked like...

A GitHub repository catalogs 475 Claude Opus 5.5 video prompts, linking creator posts, implementation tags and live remakes for practical comparison.

I Spent Week 1 of Adding AI to My ERP Deciding What the AI Is Not Allowed to Do SokoFlow...

In our previous post, we compared ALTK-Evolve with ACE and showed that how you deliver an agent's...

On September 8, 2026, Inception Labs announced Mercury 2.5 — which the company describes as the...

Following one question through the MCP stack — from the user's prompt to the final answer Modern AI...

Jev is closed and waitlisted. Kev and Laya are the open source Jev alternatives worth running. Here is how they benchmark, and where they lose.

Fine-tuning Kev-0.8B with LoRA, SFT and ReST on one T4 to detect outdated facts: the data, the training run, the test sets and how it compares with hosted LLMs.

Every API call you make to an LLM is secretly two jobs glued together. The first reads your entire...