https://www.iqshard.com/dev-log What got built, what got decided, and what is still unresolved. en Wed, 23 Sep 2026 16:29:45 GMT https://www.iqshard.com/dev-log/kimi-k3-vs-glm-5-3 https://www.iqshard.com/dev-log/kimi-k3-vs-glm-5-3 Tue, 25 Aug 2026 00:00:00 GMT Scorecards give Kimi K3 the edge over GLM-5.3. Matched metrics, independent evals, and finished-task costs tell a more useful story for anyone building software. https://www.iqshard.com/dev-log/glm-5-3 https://www.iqshard.com/dev-log/glm-5-3 Sat, 15 Aug 2026 00:00:00 GMT GLM-5.3 is a post-training-only upgrade that nearly doubles AutomationBench, more than doubles ExploitBench, and closes the gap to GPT-5.6 Sol on agentic coding. https://www.iqshard.com/dev-log/better-context-not-more-context https://www.iqshard.com/dev-log/better-context-not-more-context Thu, 23 Jul 2026 00:00:00 GMT A large context window tells you how much a model can accept, not how much it can use well. Better answers come from smaller, targeted context https://www.iqshard.com/dev-log/kimi-k3 https://www.iqshard.com/dev-log/kimi-k3 Fri, 17 Jul 2026 00:00:00 GMT Moonshot AI releases Kimi K3 with full open weights — a 2.8T-parameter MoE with native vision and a 1M-token context, and the closest an open model has come to the closed frontier. https://www.iqshard.com/dev-log/agent-vs-chatbot-cost https://www.iqshard.com/dev-log/agent-vs-chatbot-cost Fri, 19 Jun 2026 00:00:00 GMT A chatbot and an agent run on the same kind of model and the same tools. What makes an agent burn far more tokens is not how it works, but what it is for https://www.iqshard.com/dev-log/glm-5-2 https://www.iqshard.com/dev-log/glm-5-2 Wed, 17 Jun 2026 00:00:00 GMT GLM-5.2 holds a stable 1M-token context and leads open-weights models on long-horizon, tool-using work — the kind of work that resembles a real job. https://www.iqshard.com/dev-log/token-types https://www.iqshard.com/dev-log/token-types Tue, 02 Jun 2026 00:00:00 GMT A request is made of three kinds of token — input, output, and cached — and each is handled differently. How they work, and why caching reuses text across users https://www.iqshard.com/dev-log/leaky-bucket https://www.iqshard.com/dev-log/leaky-bucket Mon, 25 May 2026 00:00:00 GMT Three rate-limiter designs apply the same leaky-bucket concept as a queue, an activity meter, or a replenishing token budget. https://www.iqshard.com/dev-log/what-are-tokens https://www.iqshard.com/dev-log/what-are-tokens Fri, 22 May 2026 00:00:00 GMT Tokens are how AI models read and write, and how you are charged. A plain guide to input, output, and cached tokens for people who sign the invoices https://www.iqshard.com/dev-log/dgx-b300 https://www.iqshard.com/dev-log/dgx-b300 Wed, 29 Jul 2026 00:00:00 GMT Eight Blackwell Ultra GPUs, 2.304 TB of HBM3e, and NVLink stitching eight accelerators into one very large GPU — inside the NVIDIA DGX B300.