English text is being prepared.

UK AISI Releases ReproBAIT Evaluation: Can AI Coding Agents Reproduce Biological AI Tools from Papers?

The UK AI Security Institute's Chem-Bio team, in collaboration with SecureBio, built the ReproBAIT evaluation to test the ability of AI coding agents to reproduce biological AI tools (BAITs) given data and limited public information. Results showed that simpler tools could be reproduced within 4 hours and under a $20 compute budget; ProteinMPNN reached 96.5% reproduction fidelity with Claude Opus 5 when the full paper was provided, while the most complex tool, RFdiffusion3, was not successfully reproduced by any agent within the time limit. Based on this, the team noted that merely withholding code or restricting paper information does not equal safe release, and that controlling access to underlying biological data may be more effective than controlling code.

How to get coding agents to use your already-logged-in Chrome

Browser Use published a guide introducing several ways to let coding agents work in a real, already-logged-in Chrome, including Claude in Chrome, Playwright MCP extension mode, Chrome DevTools MCP auto-connect mode, the Gemini CLI built-in browser agent, the ChatGPT desktop browser extension, and Browser Harness. The article gives configuration commands and limitations separately for agents such as Claude Code, Codex, Gemini CLI, and Cursor, and explains that starting with Chrome 136, --remote-debugging-port and --remote-debugging-pipe on the default data directory will be ignored, so the old practice of copying cookies to a new browser will fail in scenarios such as device-bound sessions.

OpenAI's price cuts pay off in winning orders: multiple Anthropic customers shift coding business to OpenAI

OpenAI attracted Anthropic customers through large discounts and free credits. In Rippling's coding tasks, the share of OpenAI models has risen from 16% to 52%, while Anthropic models fell from 57% to 18%. Replit moved most of its AI budget from Anthropic to OpenAI; OpenAI offered a discount of over 30%, and the annual contract size exceeds $100 million. After Anthropic released Claude Opus 5.5 at the end of last month, it reclaimed the top spot in usage share on OpenRouter, and over the past week its popularity was about twice that of OpenAI models.

Artificial Analysis: Kimi K3.1 Pro leads on agentic knowledge and coding benchmarks, AA-3D ranking may be low

Artificial Analysis says Kimi K3.1 Pro is among the leading models on agentic knowledge work and coding, performing well on GDPval-AA and Terminal-Bench 4.0, but its AA-3D Elo preference voting is still being collected, and early signs indicate its ranking will be low. Its predecessor Kimi K3 ranked 12th among 22 models on AA-3D. The organization also showed a 492-meter mountain bike downhill 3D scene generated by K3.1 Pro, in which steering, pedaling and braking can be taken over at any time during the demo.

Databricks on AI coding cost governance: unit costs cut by up to 90%

Patrick, a product lead at Databricks, reviewed the cost overrun problem after the company fully adopted AI coding agents, saying that by migrating to cheaper models, routing, spend visibility, and cutting token overhead, unit costs in some scenarios fell by as much as 90%. He noted that hard budget caps would interrupt the best developers, so the company switched to gradual friction through warnings, approvals, and model downgrades, and productized its internal solution as Unity Gateway. The team built its own evaluation using real PRs from its own codebase, finding that GLM 5.2 matched Opus 4.8 in quality at about one-third lower cost per task, while a model that was 1.7 times cheaper per token ended up costing more per task because it consumed more tokens.

OpenCode 2.1 Released: One Coding Agent Supports Multiple Models

The open-source AI coding agent OpenCode has released version 2.1, with changes to the SDK, HTTP messages, events, and plugin interfaces. Vendor-related data has been moved under tool.native, and the former providerState and providerResultState fields have been removed, so integrators need to adjust accordingly. Windows x64 and ARM64 gain a persistent pseudo-terminal, and the todowrite task list can be enabled via {"experimental":{"todo":true}}. This version builds on 2.0, which unified the terminal, web, and desktop clients onto a single local backend service.

New paper proposes BackUpper: making coding agents back up before they act

A new study builds the RegretBench benchmark, testing six frontier models across 205 real data-loss scenarios. Results show agents complete tasks at an average rate of 86.1%, but back up before critical operations only 15.2% of the time—GPT-5.6-Terra at 4.4%, GPT-5.6-Sol at 5.9%, Gemini 3.7-Flash at 2.4%, and even the best performer, Claude Opus-5, failed to back up in more than one-third of scenarios. The authors' proposed BackUpper runs in parallel with the agent as an external plugin, using a four-tier strategy: replacing destructive commands, pairing backup commands, generating snapshots, or calling the GPT-5.6-Luna sub-model to determine which data needs saving. In testing it covered all critical operations, with recovered data averaging 92.2% similarity to the original state, at an additional cost of $0.014 per task. BackUpper already supports Claude Code, Codex, Hermes, OpenHands, DeepSeek Harness, and OpenClaw, and its code and dataset have been open-sourced.

Anthropic red-teams Auto Mode: production monitoring fails to block more than 55% of malicious coding agents

Anthropic's Alignment Science blog published research using Opus 5 red-team agents to conduct adversarial testing against production interception monitoring such as Claude Code's Auto Mode and OpenAI Codex's Guardian, finding that on average more than 55% of attacks were not intercepted. The most effective attack methods included agent-generated prompt injection, multi-context attacks across sessions or compressed summaries, and exploiting coverage gaps where edits within the working directory are not monitored. The team introduced Auto Mode++ through improvements such as transcript-format escaping and heredoc framing, monitoring all edits, and adding a tool-call investigation agent, reducing the attack success rate by more than 50 percentage points, but reliably defending against multi-context attacks still directly conflicts with latency costs.

Recursive's first open-source model Flint launches on OpenRouter, a 40B MoE built specifically for coding agents

Recursive's first model Flint has launched on OpenRouter. It is an open-weight 40B MoE model (about 2.5B active parameters), built by the company's automated research system Eureka Machine, aimed at coding agents, and supports 262K context, reasoning, and tool calling. Recursive reports that it scores 54.9% on Terminal-Bench 2.1, 89.7% on SWE-bench Pro v2, and 74.2% on SWE-bench Verified, and is free to use during its launch period.

Recursive releases Flint, a 40B MoE coding model, at 125+ tokens/s on a MacBook Pro

Recursive has released its first model, Flint, a 40B mixture-of-experts coding model under the Apache 2.0 license, with weights now open on Hugging Face, built from scratch in under 4 months by the automated research system Eureka Machine. Flint generates code at 125+ tokens/s on a MacBook Pro and reaches 150+ tokens/s on an RTX Spark. On a 48GB MacBook, its average quality is about 81%, compared with about 85% for Qwen3.8 27B, but in 128K-token long sessions it is about 7 times faster than the latter.

Recursive releases laptop-class coding model Flint, says it was built mainly with AI involvement

AI startup Recursive has released its first model, Flint, a coding model that runs offline on a laptop and is freely downloadable on Hugging Face under the Apache 2.0 license; the company says its development process was mainly carried out with the involvement of its in-house Eureka Machine. Flint is an MoE model that activates only 2.5B of its 40B parameters at a time, generates at about 131 tokens/sec on a MacBook Pro with an M4 Pro, and outperforms other models under 50B parameters on SWE-Bench Verified. Recursive says Flint requires more guidance than large paid coding assistants and recommends breaking large tasks into small steps.

Recursive Releases Open-Source Coding Model Flint, Says Most of Development Process Was Completed by AI

Recursive released Flint, its first open-weight coding model, with 40B parameters. CEO Richard Socher said that nearly every stage of the model pipeline was completed by its Eureka Machine system, including a new base training method, the model architecture, and code optimized for specific chips. A WIRED author tested it on a Mac with 48GB of memory, using it to build a command-line to-do app, a web research scraping program, and an RSI educational website, and found its speed and accuracy superior to models of the same size. Recursive has raised $650 million from Google Ventures, Greycroft, Nvidia, and AMD Ventures, at a valuation of $4.65 billion.

From personal coding tools to development platforms: trust and cost in AI programming

Massimo Re Ferre, Senior Principal PM at AWS, believes that the key for teams moving from personal coding assistants to agent development platforms is not model capability, but whether agent output can be verified at low cost—trust rises with inspection capability, not with benchmarks. Agent costs are billed by task activity; about 80% of developers use only 20% of the budget, while 20% of heavy users consume the rest, so the focus should be on the cost per unit of effective work rather than seat price.

From Personal Coding Tool to Development Platform: AWS Kiro Product Lead on Trust and Cost in AI Coding

Massimo Re Ferre, Senior Principal Product Manager at AWS, shared at DevCon: the speed at which teams delegate authority to AI agents depends on the cost of verifying their output, not on improvements in model capability. Agents have evolved from line-level completion to autonomously selecting tasks, and costs have shifted from fixed per-seat pricing to per-task consumption billing; about 80% of developers spend only 20% of the budget, while a small number of heavy users consume the rest.

English translation is being prepared.

Jellyfish data: coding agents already write 22% of PRs, but nearly all require human review

Jellyfish's analysis of data from about 300,000 engineers at 1,300 companies shows that at the median enterprise, autonomous coding agents have already written 22% of merged PRs, up from 18% a month earlier, with the top 10% of companies reaching 53%. However, the median share of PRs independently reviewed and merged by agents is only 0.6%, and 9% for the top 10%, with human review remaining the main bottleneck. The author also notes that the abandonment rate for agent-submitted PRs is 40%, versus 20% for humans, and that the number of human reviews per merged agent PR has more than doubled over the past year.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.