The Agent SDK is now available. Define agents as code, with prompts in Markdown and tools in TypeScript.
Define agents as code, with prompts in Markdown and tools in TypeScript. Deploy them on its hosted infrastructure, or run them on your own. Built-in evals let you measure every change and make your agents better over time.
GPT-6.1 Terra, OpenAI's newest model, is now generally available in GitHub Copilot.
It is designed for everyday agentic coding and terminal work at a lower cost than GPT-6.1 Sol.
03Updates
What's new in Gemini 4 Argon | Gemini API
Gemini 4 Argon ( gemini-4-argon ) is its latest, most capable model.
It can reason longer on hard problems for better results on complex coding, agentic, math, and science tasks.
04Updates
OpenAI: GPT-6.1 Terra: delivering fast and capable computer use at a lower price.
GPT-6.1 Terra: delivering fast and capable computer use at a lower price.
GPT-6.1 Terra understands what's on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it.
05Updates
AWS: OpenAI GPT-6.1 Terra is now generally available on Amazon Bedrock
Today, AWS announces the general availability of OpenAI GPT-6.1 Terra on Amazon Bedrock.
GPT-6.1 Terra expands the GPT-6 family with a balance of capability, speed, and cost efficiency for everyday workloads.
06Updates
arXiv: Governing Enterprise AI Traffic at Scale: Design and Operation of a Policy-Driven, Multi-Vendor LLM Gateway on Envoy
Large organizations consume large language models (LLMs) from several vendors at once, each with its own API dialect, credential model, streaming protocol and pricing.
The arXiv Coding-Agent Research page documents the arXiv update.
07Updates
AWS: Amazon WorkSpaces Core Managed Instances now supports ModifyVolume API
Amazon WorkSpaces Core Managed Instances now supports the ModifyVolume API, which lets you change volume size, volume type, IOPS, or throughput in place, without detaching the volume or restarting the instance.
With ModifyVolume, you can increase volume size up to 16,384 GiB, switch between supported Amazon EBS volume types, adjust IOPS for gp3, io1, and io2 volumes, and modify throughput for gp3 volumes.
Over 600 million people in sub-Saharan Africa lack electricity, relying on costly and harmful fuel sources.
To bridge this gap, Google partnered with Ignite Energy Access to solve complex logistical challenges, using satellite imagery and advanced data models to map remote homes and optimize solar hardware delivery.
11Updates
Working with academic communities to build scientific understanding
OpenAI shares model results on the Kakeya conjecture, Exponential Time Hypothesis, and quantum chromodynamics-and how researchers can verify and build on them.
The OpenAI News page documents the OpenAI update.
12Updates
Concerning behavior from a popular open-weight AI model
For nearly two years, one comforting idea has shaped how companies and government contractors adopt open-weight AI models that anyone can download and run.
The Reference discovery page documents the Cohere update.
Cognition: Introducing the new Devin Security: powered by Claude Mythos and frontier cyber models, plus hundreds of millions of dollars of...
Introducing the new Devin Security: powered by Claude Mythos and frontier cyber models, plus hundreds of millions of dollars of credits to help teams find, prioritize, and fix urgent vulnerabilities.
The X @cognition page documents the Cognition update.
Infinity Machine: a self-improving AI agent for research
The harness itself(prompts, tools, skills, playbook, meta agent) is an editable genome, changed only through a kernel-enforced gate with held-out and cost checks.
The arXiv Coding-Agent Research page documents the arXiv update.
Stealing OpenAI Reasoning Traces via Compaction Smuggling
Reasoning models like gpt-5.6-sol think before they answer, but OpenAI only shows API users optional summaries of that chain of thought.
One reason is to prevent distillation (a process where a smaller "student" model is trained on a larger "teacher" model's outputs).
19Updates
Discover 1000+ services on Vercel Connect from the CLI
Results can be filtered by category, connection method (such as mcp or oauth ) or the product a service exposes.
You can also limit results to services that forward webhook triggers.
20Updates
MiRL: Error-Targeted Reinforcement Learning for Correcting Terminal Agent Misbehaviors
Existing error-targeted methods are designed for single-turn tasks, and existing analyses of agent failures are used to evaluate and debug agents rather than to train them.
The arXiv Coding-Agent Research page documents the arXiv update.
arXiv: Quantifying Claim Smoothing by LLM Research Agents
Deep research, or any use of large language models (LLMs) to read a large document collection and produce insights, requires synthesizing information from different sources.
The arXiv Language-Model Research page documents the arXiv update.
24Updates
AWS: Fleet-wide computer vision lifecycle at the edge with SageMaker SDK v3
This post updates the pattern introduced in the 2023 series Build an end-to-end MLOps pipeline for visual quality inspection at the edge for the current Amazon SageMaker Python SDK v3.
The AWS Machine Learning Blog page documents the AWS update.
EvoRouter: Cost-Efficient Multi-Agent Routing via Insight-Guided On-Policy Distillation
However, the resulting configuration space grows combinatorially, while sparse execution rewards provide limited fine-grained supervision for interdependent planning decisions.
The arXiv Coding-Agent Research page documents the arXiv update.
Phronix: The Capability-Centric Learning Paradigm for Agents
Scientific computing increasingly relies on large language model-based agents for performance engineering tasks such as instrumentation, analysis, and workflow porting.
The arXiv coding-agent research feed release records 2610.31789v1 for arXiv.
28Updates
arXiv: The Agent in the Middle: How AI-Mediated Access Reshapes Web Vulnerability Exploitation
This shift changes how existing Web vulnerabilities can be exploited.
The arXiv Coding-Agent Research page documents the arXiv update.
Beyond On-Policy vs. Off-Policy: At What Level Should LLM Agent Distillation Follow the Student?
Distilling interactive LLM agents requires choosing between teacher-generated demonstrations and student-generated experience, corresponding broadly to off-policy and on-policy distillation.
The arXiv coding-agent research feed release records 2610.31918v1 for arXiv.
NVIDIA and Meta to Expand Collaboration on Muse and Muse Family of Models Amid Robust Growth
Meta to expand NVIDIA Vera Rubin and Vera CPU deployment to boost performance and scale workloads with NVIDIA Spectrum-X Ethernet networking and BlueField-4.
The NVIDIA Newsroom page documents the NVIDIA update.
arXiv: Profiling a Frontier Agent with EBench: From Direct Control to Harness with Policy
Frontier multimodal large language models (MLLMs) can control robots without robot-specific training, but how their capabilities compare and interact with post-trained vision-language-action (VLA) and world action model (WAM) policies across manipulation demands remains to be thoroughly evaluated.
The arXiv Coding-Agent Research page documents the arXiv update.
This data gives researchers a new way to study AI's role in the economy, with new information on the work activities people bring to ChatGPT-from preparing instructional materials and developing marketing content to programming and analyzing business data.
KAPPA: Checking the Proofs in Statistics Papers with Lean 4 and Language Model Agents
Each numbered result receives one of four verdicts: verified unconditionally; verified conditionally, with every cited or unproved input stated as a named hypothesis; issue recorded, with its place in the paper and the side at fault; or unfinished, with the reason.
The arXiv Coding-Agent Research page documents the arXiv update.
arXiv: Free to Look, Free to Reason: Simple LLM Harnesses Outdiagnose Hand-Crafted Root Cause Analysis Agents
Root cause analysis (RCA) of microservice failures has attracted many LLM agents whose investigation is hand-crafted, through standard operating procedures, fixed stages and role agents.
The arXiv Coding-Agent Research page documents the arXiv update.
S3AC: Stabilized, Selective, and Scalable Actor-Critic for LLM Reasoning
Critic free methods are widely used in reinforcement learning with verifiable rewards (RLVR) for long chain of thought (CoT) reasoning, avoiding difficult and costly critic training.
The arXiv Language-Model Research page documents the arXiv update.
40Updates
Knowing Whom to Trust is not Enough: The Trust--Agreement Gap in Repeated Multi-Party LLM Negotiation
Large language model (LLM) agents increasingly participate in negotiations on behalf of individuals and organizations, yet they remain vulnerable to manipulation, particularly in repeated multi-party interaction.
The arXiv coding-agent research feed release records 2610.31971v1 for arXiv.
41Updates
arXiv: A Unified Arena for LLM Agents across Diverse Eurogame
Interactive environments offer a practical way to evaluate large language model (LLM) agents, and competitive games extend this evaluation to decisions shaped by competing goals and other agents' choices.
The arXiv Coding-Agent Research page documents the arXiv update.
42Updates
arXiv: Preserve to Expand: Source-Aware Adaptation and Model Merging for Multilingual LLM-Based ASR
Expanding multilingual ASR systems to new languages is challenging when adaptation must preserve previously supported languages without replaying source-language training data.
The arXiv Language-Model Research page documents the arXiv update.
The NPM Remotion Renderer release records 4.0.539 for Remotion.
47Updates
How Ultrafast cut incident investigation from hours to minutes at OpenAI
These tools are being used internally, at OpenAI, and are shared here as illustrative examples of how frontier AI is supporting use cases across its teams.
Even a carefully proofread paper can leave small gaps in its proofs or its stated assumptions. We introduce PaperFormalizer
arXiv coding-agent research feed - arxiv.org · 2026-10-27
UNVERIFIED — supplemental reports
UNVERIFIED Non-official information. Unverified report. Reuters reports: Cadence unveils suite of AI models that chipmakers can run in-house. The claim from Reuters is not independently confirmed. Confirmation to watch for the Reuters report: A first-party page from Reuters that states it. Reuters 2026-10-26T17:22:18Z
UNVERIFIED Non-official information. Unverified report. TechCrunch reports: Alexander Embiricos, the OpenAI executive responsible for dots, never wants to have a meeting with someone else's AI agent. The claim from TechCrunch is not independently confirmed. Confirmation to watch for the TechCrunch report: A first-party page from OpenAI that states it. TechCrunch 2026-10-27T02:00:00Z
UNVERIFIED Non-official information. Unverified report. The Verge reports: OpenAI announces three big science discoveries and a new way to work with academics. The claim from The Verge is not independently confirmed. Confirmation to watch for the The Verge report: A first-party page from OpenAI that states it. The Verge 2026-10-26T23:34:58Z