i14 Labs 发布 Sugar Glider 架构：固定 KV cache 支撑长上下文智能体
i14 Labs 新发的 Sugar Glider 把 KV cache 做成固定大小，100 万 token 上下文下单机吞吐能到基线的 32 倍，部署时还能调记忆大小不用重训。做长跑 agent 的开发者值得看，尤其是一直被上下文裁剪和推理成本折磨的团队。
i14 Labs 推出 Sugar Glider 长上下文 LLM 架构家族，核心特性是随上下文增长保持固定大小的 KV cache，实现每 token O(1) 推理。官方数据显示，在 100 万 token 上下文下，单台服务器可承载的请求数最高达到基线模型的 32 倍。
架构将推理时的内存占用与上下文长度解耦，开发者可在部署阶段直接选择记忆大小而无需重新训练模型。在检索质量方面，Sugar Glider 在 10 亿 token 上下文中达到 100% 单针检索和 90.5% 的 multikey 1 成绩，且官方称已在约 1T 参数的大模型上验证了这套效率特性。
团队还表示，可将流行的开源权重模型转换为 O(1) 推理和固定内存形态，单次转换只需数小时。对需要长历史记录的 agent 应用而言，这意味着不再频繁裁剪或摘要上下文，推理服务成本结构也随之改变。
I'm bullish on new architectures for long-running agents. But long context needs to be practical to serve. Sugar Glider is a new family of LLM architectures from @i14labs built for long contexts. It keeps a fixed-size KV cache as context grows. At one million tokens, i14 reports up to 32x more requests per server than its baseline model. What does this mean in practice? Your agents could keep much longer histories without constantly trimming or summarizing them. And you can choose that memory size at deployment without retraining. Tom Mcleod @todomcl Frontier performance at a fraction of the compute. Today at @i14labs we’re launching Sugar Glider, our new family of LLM architectures! Join the waitlist for early access: i14.ai Sugar Glider delivers a new Pareto frontier: extreme memory and compute efficiency whilst maintaining frontier performance, even for large models (tested up to ~1T params). O(1) per token inference → ✅ Fixed size KV cache → ✅ Frontier long-context retrieval quality → ✅ 1 billion token long context retrieval → ☑️** (**100% -> single needle @ 1B-token context, 90.5% multikey 1 @ 1B-token context.) How did we do it? Turns out a few decades in the trenches of human memory and language model research is a pretty handy starting point to solve the “intelligence vs efficiency vs memory” conundrum. ... and just for something extra, we’ve been converting some of the most powerful and popular open-weight models to O(1) inference and fixed memory, with each conversion taking just hours :) Tune in via our waitlist for early model access i14.ai More results dropping soon!
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Sugar Glider, i14 Labs, 长上下文, KV cache.
Source attribution: curated from elvis (@omarsar0).
