Akhilesh Balasingam, Amadou Ngom, Geoffrey X. Yu, Tim Kraska
Recent work has shown that LLMs can synthesize highly instance-optimized database systems for fixed workloads, but to date these systems only replace the DBMS's native storage and assume static data: users must migrate their data, often give up ad-hoc queries, lose acceleration once the data changes, and lack formal correctness guarantees. We present JetStream, which instead generates query-specific accelerators that extend an existing DBMS. Given a set of parameterized query templates, a staged agentic workflow synthesizes accelerators that run directly over the DBMS's native tables and may maintain custom auxiliary state, while the host DBMS remains responsible for storage, transactions, recovery, and all the other queries. A fixed, engine-neutral substrate keeps auxiliary state transactionally consistent under writes, while generated backend adapters handle engine-specific hooks. JetStream generates two kinds of accelerators: tested accelerators, which may use any optimization and are validated by differential testing, and verified accelerators, which trade some performance for a machine-checked proof that every answer satisfies a trusted specification derived from the SQL. On TPC-H at SF=20, JetStream achieves a $584\times$ geomean read speedup over DuckDB ($89\times$ verified) and retains $139\times$ ($46\times$ verified) under a continuous update stream of 10,000 modified rows per second.
Mohammad Javad Amiri, Carsten Binnig, Andreas Kipf, Tim Kraska, Ryan Marcus, Johannes Wehrstein, Chenyuan Wu, Geoffrey X. Yu 等 9 位作者
For the first time in history, high-performance code is cheap. Today's coding agents can generate bespoke, instance-optimized data systems with excellent performance, and such agents will continue to improve as LLMs advance. But a purely agentic approach, even when validated with strong testing techniques, does not automatically inherit the correctness guarantees our community has spent decades building. We propose a middle ground between hand-crafted and agent-crafted systems: correctness firewalls, lightweight design patterns under which a synthesized component can affect a system's performance but never its results. Behind a correctness firewall, agents are free to synthesize fast components while the overall system remains correct. We present three case studies, each demonstrating a different type of firewall: cardinality estimation (advisory), searching sorted data (guarded), and data compression (certified). In early experiments, synthesized components search sorted data up to 3x faster than human-designed learned indexes and can compress text 1.5x smaller than Zstd. Drawing on these case studies, we outline a research agenda for how data management systems can exploit coding agents without giving up their hard-won correctness guarantees, from protecting synthesized code to measuring the cost of correctness.
AI 速览：提出递归语言模型（RLM）推理范式，让 LLM 把超长提示当作外部环境，通过编程式检查、分解并递归调用自身来处理远超上下文窗口的输入。
We study allowing large language models (LLMs) to process arbitrarily long prompts through the lens of inference-time scaling. We propose Recursive Language Models (RLMs), a general inference paradigm that treats long prompts as part of an external environment and allows the LLM to programmatically examine, decompose, and recursively call itself over snippets of the prompt. We find that RLMs can successfully process inputs up to two orders of magnitude beyond model context windows and, even for shorter prompts, dramatically outperform the quality of vanilla frontier LLMs and common long-context and coding scaffolds (e.g., on GPT-5 by a median across the evaluated benchmarks of $26\%$ against compaction, $130\%$ against CodeAct with sub-calls, and $13\%$ against Claude Code) across four diverse long-context tasks while having comparable cost. At a small scale, we post-train the first model around the RLM. Our model, RLM-Qwen3-8B, outperforms the underlying Qwen3-8B model by $28.3\%$ on average and even approaches the quality of vanilla GPT-5 on three long-context tasks. Code is available at https://github.com/alexzhang13/rlm.