AITOPAI热报
Products

Reason 发布：号称以 20 倍更低成本跑出前沿级编码性能

elvis@omarsar0
Post
View the post
Why we picked it

一个叫 Reason 的编码 agent harness，号称成本降到前沿基线的零头，还能在 Terminal-Bench 和 DeepSWE 上跑赢主流方案。天天跑 agent 被账单吓到的开发者可以先去领个早期访问免费试试。

Adi Singh 宣布推出 Reason，定位为成本效率最高的编码 agent harness，声称能以前沿基线一小部分的成本完成任务，成本最多降低 20 倍。

Body

在基准表现上，Reason 称在 Terminal-Bench 和 DeepSWE 上超过主流 harness，同时把每个任务的成本压到远低于前沿模型基线的水平。Omar Sansevero 转发评价称，harness 通常难以兼顾效率与性能，Reason 做到了两者兼得，应用本身完成度也不错。

对开发者而言，这类 harness 的价值在于控制 agent 跑任务的账单开销——目前编码 agent 的失控开销（runaway bills）是实际使用中的常见痛点。产品当前提供早期访问和免费算力额度，可到 reasonmachines.com 免费试用。

Original · elvis

Very impressed by this. In the harness world, you often don't get the performance you need when you try to maximize efficiency and speed. Looks like Reason is unlocking both. The app is also well-built. Worth a try. Adi Singh @xadisingh Frontier coding performance at up to 20x lower cost. Introducing the world’s most cost-efficient harness: Reason Stop burning thousands on runaway agent bills. Reason outperforms top harnesses on Terminal-Bench and DeepSWE while dropping cost per task to a fraction of the frontier baseline. Comment “Reason” and we’ll send you early access + free compute credits. Try for free: reasonmachines.com

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Products" 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 Reason, Terminal-Bench, DeepSWE, coding assistants.

Source attribution: curated from elvis (@omarsar0).

#Reason#Terminal-Bench#DeepSWE#coding assistants#agents

Keep reading

Recent stories with the same tag or section
ProductsBuilders01:24
Chinese contentReason 发布低成本智能体 harness，宣称成本降至前沿基线的 1/20

做智能体编程的可以看看这个叫 Reason 的 harness，主打省钱：官方说在 Terminal-Bench 和 DeepSWE 上跑赢同类，单任务成本能压到前沿基线的一小部分。如果你正被 agent 的 API 账单烧得心疼，可以申请早期免费额度试试。

#Reason#Terminal-Bench#DeepSWE#coding assistants
AI Breakfast@AiBreakfastSource
ProductsBuildersSelected05:43
Browser Use cuts agent cost 33% by fixing prompt cache hit rate

Cache hit rate is often invisible in agent eval reports, yet here it accounted for a third of the cost per task with zero score change, a lever most teams never instrument. Engineers building or benchmarking browser agents can audit their own harnesses for cache-breaking behaviour and adopt browser-use 1.0.4 directly. The standardised comparison against six browser tools inside Codex gives buyers a concrete reference point when choosing a harness.

#Browser Use#cache optimization#coding assistants
Browser Use@browser_useSource
ProductsBuildersMulti-source05:24
Chinese content美国 Gemini 用户现可绑定 Google Wallet 并接入银行账户

Google 这次把 Gemini 接上了真实钱包和银行数据，靠 Plaid 读信用卡和账户记录。比起 Apple Intelligence 还停留在语音和照片整理，Gemini 已经能分析你的实际消费了。如果你在美国用 Gemini，可以在设置里试试绑定，顺便留意一下授权范围。

#Gemini#Google Wallet#Plaid#multimodal
Event
TestingCatalog@testingcatalogSource
ProductsBuildersMulti-sourceSelected05:19
Artificial Analysis launches Diarization Index, NVIDIA leads

A standard WDER metric shifts how transcription APIs will be evaluated, since many buyers currently judge quality only on word accuracy and miss attribution failures that break meeting summaries and call analytics. Teams building voice agents, meeting assistants or call-centre tooling can use the 97-conversation suite to benchmark vendors before committing. NVIDIA's Nemotron 3.5 Transcribe at 3.3% versus Voxtral Small Transcribe 3 at 7.5% sets a clear quality bar for providers to chase.

#Nemotron#Voxtral#NVIDIA#speech recognition
Event
Official accountArtificial Analysis@ArtificialAnlysSource
ProductsBuildersOfficialSelected05:00
Google opens Antigravity /teamwork multi-agent tools to external users

The real signal is that Google moved a multi-agent collaboration framework from internal demos to external availability, so claims about agents solving open research problems can now be independently tested. Researchers in mathematics, chemistry, systems software and ML engineering can apply the toolset to their own unsolved tasks. Teams comparing agent frameworks should check the linked results before committing. Availability details and pricing for ASEAN and global users were not stated in the announcement.

#Antigravity#Google#agents#MCP and tools
Primary sourceGoogle Research@GoogleResearchSource
ProductsBuildersMulti-sourceSelected71°04:55
llama.cpp v1.0.0 ships as first stable release

A stable 1.0 changes llama.cpp from an actively shifting project into infrastructure teams can standardise on, which matters for anyone deploying models on-premise or offline. Cohere's prompt adoption shows model vendors now treat llama.cpp support as a launch requirement, a signal worth tracking for teams choosing which architectures to build against. Developers embedding local inference in apps can treat the v1 API as a stable contract rather than pinning versions defensively.

#llama.cpp#Cohere#Hugging Face#local deployment
Event
Official accountCohere@cohereSource