Back to Tool
M

Maxwell

Tool Oct 19, 2026

doubleAI's performance engineering system, presented as a Code Execution Model that inspects running programs and proposes hardware-specific optimizations.

Visit primary link 2 sources attached
doubleAI's component chart compares LTX-2.5 at 249.4 seconds with its optimized pipeline at 133.9 seconds on two B200 GPUs.
doubleAI's October 19 author-published timing breakdown for one five-second 4K clip on two B200s. The full chart is preserved; these are attributed workload measurements, not AITrending tests or a universal speedup. Source: doubleAI / Maxwell LTX-2.5 collaboration report
ai-agentsmultimodal

What happened lately

Maxwell timeline

Maxwell is doubleAI’s performance engineering system, marketed as a Code Execution Model. The product page describes using execution traces, hardware information and runtime data to improve programs without editing their source, with formal equivalence checks for runtime replacements. Early access is offered through the website; the reviewed product page does not establish general pricing or a public installation command.

In a separate October 19, 2026, LTX-2.5 collaboration report, doubleAI describes algorithmic and GPU-kernel changes that reduce a five-second 4K clip’s generation time from about 250 to 134 seconds on two B200 GPUs. The report says the soundtrack is bit-identical and reports 41 dB video PSNR on LTX’s benchmark prompt. That is a vendor-reported workload result, not proof of bit-identical video or a universal speedup. AITrending has not reproduced it.