Agents

AIAndrej Karpathy's year-in-review lists the LLM paradigm changes he found most notable in 2025. He highlights Reinforcement Learning from Verifiable Rewards (RLVR), which drove most capability gains as labs ran longer RL training, and describes LLM intelligence as jagged, strong in verifiable domains and weak elsewhere. He also covers Cursor-style LLM apps, Claude Code running on the user's computer, vibe coding, and the case for a visual LLM GUI.

Why it matters: Karpathy ties the year's shifts to RLVR, jagged capability, and local agents, giving readers a framework for judging how LLM progress is changing.

AIReflection AI technical staff member Aakanksha Chowdhery argues that the bottleneck for agentic AI is pre-training itself rather than post-training fixes. Drawing on her work on PaLM and early Gemini, she says next-token prediction breaks down for long-horizon planning and that objectives, attention, and training data must evolve.

AIXiaomi released and open-sourced MiMo-V2-Flash, a Mixture-of-Experts model with 309B total and 15B active parameters, under the MIT license. The company reports 73.4% on SWE-Bench Verified, the top score among open-source models, and inference at 150 tokens per second for $0.1 per million input tokens and $0.3 per million output tokens. It supports a hybrid thinking mode and a 256k context window.

Why it matters: The post gives architecture, speculative decoding speedup, and pricing figures, which help readers judge how the efficiency claims are achieved and what they cost.

AIRunway announced GWM-1, its first general world model family, built on Gen-4.5 and generating frames autoregressively in real time under interactive control. It comes in three variants: GWM Worlds for explorable environments, GWM Avatars for conversational characters, and GWM Robotics for robotic manipulation. Runway also says it is working toward unifying these domains under a single base world model, and GWM Robotics includes a Python SDK.

Why it matters: The post separates three GWM-1 variants and ties each to a concrete use, which clarifies where a general world model would fit compared with a single model.

AIThe main post argues that AI has moved the bottleneck from execution to human judgment, vision, taste, and context. AI can explore options but cannot determine which is right, so specialization now lies in judgment rather than execution. The background post, by designer @ryolu_, adds that small teams with overlapping skills may outperform larger specialist teams coordinating handoffs.

AICognition's 2025 performance review says Devin works best on clear, verifiable tasks such as migrations, vulnerability fixes, and unit tests. The company reports a 67% PR merge rate, up from 34% last year, and cites a bank that cut migration time per file from 30-40 hours to 3-4 hours. It also says Devin struggles with ambiguous requirements, mid-task scope changes, and soft-skill work that still needs human engineers.

Why it matters: The report pairs concrete migration, vulnerability, and test-coverage figures with named weaknesses, letting engineering leaders judge where an agent fits in their own workflow.

AIAman Sanger of Cursor says the team is hard at work on Composer 2. The post gives no features, benchmarks, or release timing, though it follows Cursor's $2.3B Series D raise and over $1B in annualized revenue.

AIMoonshot AI released Kimi K2 Thinking, an open-source thinking model that interleaves step-by-step reasoning with tool calls across 200 to 300 sequential invocations. The model is a 1T-parameter mixture-of-experts with 32B activated parameters and a 256k context window, and it uses native INT4 quantization for roughly 2x faster generation. The model card reports benchmark results on HLE, BrowseComp, and other tests, and recommends vLLM, SGLang, or KTransformers for deployment.

Why it matters: The model card gives benchmark tables, quantization details, and deployment settings, letting readers compare Kimi K2 Thinking against GPT-5 and other models on specific tasks.

AIWindsurf has launched Codemaps, AI-annotated structured maps of a codebase powered by SWE-1.5 and Claude Sonnet 4.5, which users can generate from a task prompt using a Fast (SWE-1.5) or Smart (Sonnet 4.5) model. Codemaps links grouped code sections to exact lines and can be referenced in Cascade with @{codemap} to give agents more specific context.

AIChip Huyen argues that many AI product failures stem from user experience, data quality, and organizational structure rather than the AI itself. She cites a chatbot whose traction improved after adding pre-populated questions and a voice option for users whose hands were busy, and a lead scoring model that was broken because marketing wasn't asking the right questions. She also notes that senior engineers gain the most from AI coding while resisting it more, and recommends building small tools for daily frustrations to solve the "idea crisis."

AICognition has released SWE-1.5, a model optimized for software engineering that it says reaches near-frontier coding performance while running at up to 950 tok/s with Cerebras inference. The company reports it is 6x faster than Haiku 4.5 and 13x faster than Sonnet 4.5, and it is available now in Windsurf. The post's SWE-Bench Pro chart places SWE-1.5 at 40.08%, behind Sonnet 4.5 at 43.60%, and it notes that the model was trained with reinforcement learning on the Cascade agent harness.

Why it matters: The post pairs a benchmark chart with a 950 tok/s speed claim and describes how harness, RL environments, and inference were co-designed, useful context for judging the speed-versus-quality tradeoff.

AICognition says its autonomous coding agent Devin can complete a .NET Framework to .NET Core migration in as little as two weeks, using a Strangler Fig approach adapted from Jimmy Bogard's guide. The post says Devin handles planning via Ask Devin and DeepWiki, dependency sharing, controller and view conversion, and session state adaptation through a remote app.

AIFactory has announced a partnership with Amazon Web Services and made its Droids agent platform available on the AWS Marketplace. Enterprise teams can use existing AWS Enterprise Discount Program commitments to buy Factory, with Droids accessible from CLI, Terminal UI, Web, Slack, Linear, and an IDE overlay. The source cites 31× faster feature development, 96.1%+ reduction in migration times, and 95.8% reduction in incident resolution times.

AICognition introduces SWE-grep and SWE-grep-mini, fast agentic models trained with reinforcement learning for multi-turn context retrieval in coding tasks. The company says they match frontier coding models at retrieval while taking an order of magnitude less time, and they power the Fast Context subagent in Windsurf. The models issue up to 8 parallel tool calls per turn within 4 turns, and Cerebras serves SWE-grep-mini at over 2,800 tokens per second and SWE-grep at over 650 tokens per second.

Why it matters: The post explains the speed-intelligence tradeoff in agentic code search, showing how parallel tool calls and RL training change the cost of retrieving context for coding agents.

AICognition rebuilt its Devin coding agent for Claude Sonnet 4.5, reporting 2x faster performance and 12% better results on its Junior Developer Evals, now available in Agent Preview. The team found the model is aware of its context window, which led to premature wrap-up behavior that they countered with repeated prompts and a 200k usage cap within a 1M token beta.

Why it matters: The post explains which agent behaviors changed under Sonnet 4.5, such as context-window awareness and note-taking, that forced a rebuild rather than a simple model swap.

AICognition says Claude Sonnet 4.5 is available in Devin starting today, improving its planning performance by 18% and end-to-end eval scores by 12%. Cognition says the model's testing of its own code lets Devin run longer and handle harder tasks.

AICognition, maker of the AI software engineer Devin, raised over $400M at a $10.2B post-money valuation led by Founders Fund. The company says its acquisition of Windsurf more than doubled its ARR, with combined enterprise ARR up over 30% in the seven weeks after the deal. It also reports Devin ARR grew from $1M in September 2024 to $73M in June 2025, with total net burn under $20M.

AIEight Sleep integrated Cognition's Devin into its data workflows, letting staff tag Devin in Slack to query Snowflake, dbt, and Looker and check Amplitude. The company says it is now shipping 3x as many data features and investigations each week, with its ad-hoc data request queue near zero. Devin was used to trace a suspicious revenue spike to a better-than-expected email campaign.

AICognition published a guide titled "Build Your Own AI Data Analyst" on its blog, but the provided text contains only a navigation listing of unrelated posts and no guide content. No tools, models, features, or steps can be verified from the source.

AICognition says generative AI agents such as Devin can read internal docs and convert Jenkins pipelines to GitHub Actions syntax, replacing custom plugins with Actions or APIs and validating the results. The company claims enterprises can cut multi-year migration efforts to a few months, with Devin running inside the customer's secure environment so code and secrets stay internal.

AICognition's Devin is now compatible with the Model Context Protocol (MCP), letting users connect favorite MCP servers through a new MCP Marketplace found in Settings. The source cites example uses including querying Datadog and Sentry logs, creating Notion docs, Google Docs, and Linear tickets, and interacting with Figma, Airtable, Stripe, and Hubspot.

AICognition Blog published "Coding Agents 101: The Art of Actually Getting Things Done" on June 26, 2025, as a guide to using coding agents effectively. The available source text contains only navigation and a list of other Cognition posts, so no specific techniques, features, or results can be verified from it.

AICognition argues that parallel multi-agent architectures are fragile because subagents act on conflicting, unshared assumptions. It proposes two principles for reliable agents: share context and full agent traces, and treat actions as carrying implicit decisions. The post recommends simpler single-threaded designs for most cases and notes that context compression and fine-tuned models can extend long-running tasks.

Why it matters: The post explains concrete failure modes of parallel multi-agent setups and offers two context-sharing principles, useful for anyone designing long-running agent systems.

AICognition has released Devin 2.1, which reports its confidence in completing tasks using green, yellow, and red ratings. The company says green scores led to twice the likelihood of a merged PR compared with red, and Devin now also answers codebase questions and scores Linear and Jira issues.

Why it matters: The post explains how Devin now shows confidence scores and asks clarifying questions, which changes how teams can decide which tasks to hand over.

AICognition has released Devin 2.0, a new agent-native IDE experience with a flexible plan starting at $20. The update lets users run multiple parallel Devins, each with its own cloud-based IDE, and adds Interactive Planning, Devin Search, and Devin Wiki.

Why it matters: The release adds planning, codebase search, and auto-generated wikis to Devin, showing how an agent can prepare work before executing it.

AICognition's February 2025 Devin update adds parallel batch edits, beta GitLab support, and Sonnet 3.7, which Cognition says is the best model it has tested for debugging, codebase search, and agentic planning. Devin is about 2x faster than in October 2024, taking about 7.8 minutes on average to complete junior developer tasks in Cognition's internal evaluations. Other changes include copy-paste in Devin's browser and proactive feedback on suboptimal prompts.

AILinktree has used Devin, Cognition's AI software engineer, to merge roughly 100 pull requests in a month, mostly fixing customer-reported bugs and implementing small features. The engineering team also used Devin to add support for new social media platforms, launching five Devins, one per repo and PR, and later used the Devin API with a Playbook script to spawn multiple Devins for multi-repo features. The team says Devin works best on tasks an engineer could finish in a couple of hours.

AICognition's Devin can be triggered through its External API by GitHub Actions to automatically review pull requests, typically within five to ten minutes. The setup involves adding a workflow file, storing a DEVIN_API_KEY secret, and customizing the review prompt to match team conventions. Cognition recommends treating Devin as an extra reviewer rather than a replacement for human oversight, since it does not catch every bug.

AICognition's January 2025 Devin update improves its ability to find relevant files and reuse existing code in repositories, with changes rolling out to all users. It also adds enterprise accounts for centralized management of multiple organizations, audio message support in Slack, and pay-as-you-go billing after monthly ACU capacity is used, starting January 9.

AICrossmint said Devin became its top contributor to the open-source GOAT SDK during an initial trial, merging 8 pull requests versus 4 for the next contributor. Examples included a DEXScreener plugin built from a documentation URL and a harder Sui blockchain integration that needed three rounds of feedback and about an hour of human involvement. The company said its results depended on proper training, clear task context, and planned validation, not on treating Devin as superhuman.

AICognition has made Devin generally available to all engineering teams, with subscriptions starting at $500 per month. Over the past two weeks, Devin was made about 10% faster and about 10% more cost-efficient, especially for tasks requiring many code edits. The update also adds fixes for stuck or hanging sessions, more options to customize filters and Slack notifications, and larger machine settings for disk, RAM, and CPU.

AICognition is launching the Devin Open Source Initiative, giving selected open source maintainers 500 free ACUs on a Devin Teams plan as part of Devin's general availability launch. The post shows Devin contributing pull requests to projects including Anthropic's MCP Inspector, Dagger, and nanoGPT, with maintainers still reviewing the results. Devin's GitHub integration forwards PR comments and CI checks to help refine changes, though the company warns a human should still verify final quality.

AICognition is making Devin generally available to engineering teams starting at $500 a month, with no seat limits and access to its Slack integration, IDE extension, and API. The post recommends starting with small frontend bugs, first-draft PRs for backlog tasks, and targeted refactors, and shares open-source PR sessions where Devin resolved issues for projects including Anthropic MCP, Zod, and nanoGPT.

Why it matters: The post shows concrete open-source PR examples and the tasks where Devin works best, helping teams judge where an autonomous coding agent fits their workflow.

AICognition's December 2024 Devin update lets users tag Devin in Slack threads and ask it to create PRs, with Devin automatically responding to PR comments and lint failures. The update adds Repo Knowledge that Devin generates by scanning repositories, an Agency setting that makes Devin propose plans before executing complex tasks, and a REST API for structured input and output. Devin also gained faster session startup and enterprise options such as Okta single sign-on.

AICognition tested OpenAI's o1-mini and o1-preview in a simplified Devin-Base agent, comparing them with GPT-4o on its internal cognition-golden benchmark. The chart reports Devin-Base scores of 25.9% with GPT-4o, 34.6% with o1-mini, and 51.8% with o1-preview, versus 74.2% for the production Devin. The post also describes the benchmark's realistic environments, simulated users, and agent-based evaluation.

Why it matters: The post explains how Cognition evaluates coding agents with autonomous, environment-based tests, which shows how base-model swaps are measured in practice.

AICognition's Devin gained a September 2024 update with MultiDevin, which lets a manager Devin delegate work to up to 10 worker Devins, currently available on the Enterprise plan. Devin now automatically responds to comments on its pull requests, suggests Knowledge additions, and can restore earlier checkpoints, and the company reports up to an 80% reduction in time for common tasks.

AICognition tested whether its Devin AI agent could recover Windows machines stuck in the Blue Screen of Death after the CrowdStrike outage. Following a playbook of eight steps, Devin mounted the drive, deleted the faulty CrowdStrike files, and debugged a failed volume detach. The blog says the machine was confirmed bootable, and that playbooks are most useful when the same fix must be repeated across many machines.

AIDevin's June 2024 update lets users directly operate its VSCode editor, terminal, and browser, and adds Playbooks for recurring engineering tasks. It also adds Knowledge sharing, Machine Snapshots that persist dev environments, event-driven triggers from Slack, GitHub, and Linear, a Secrets manager, and tools for verifying Devin's work. Access remains limited to a waitlist with weekly invite releases.

AICognition reports that its agent Devin resolved 79 of 570 sampled SWE-bench issues, a 13.86% success rate, without being given the files to edit. The report says this exceeds the best previous unassisted baseline of 1.96% and the best assisted result of 4.80%. It also describes the adapted evaluation setup, a 45-minute runtime limit, and cases where Devin failed on multi-file edits.

Why it matters: The report explains how SWE-bench was adapted for end-to-end agent evaluation, with failure cases that clarify where the 13.86% result comes from and its limits.

AICognition introduces Devin as an AI software engineer that can plan and execute complex engineering tasks with a shell, code editor, and browser. On SWE-bench, Devin resolved 13.86% of issues end-to-end, versus a previous state-of-the-art of 1.96%, on a random 25% subset of the dataset. Devin is in early access, with access available through a waitlist.

Why it matters: The post pairs Devin's end-to-end task demos with SWE-bench results against prior models, letting readers weigh the claimed capability against the evaluation setup.