Tracker · Hacker News
AI stories rising on Hacker News this week
The AI stories that gained the most points over the last 7 days, ranked by how much they gained while SteadyVane watched. Only items the relevance reviewer scored on-topic for AI tools (6/10 or higher) are listed. Each name opens its history: every reading, day by day.
points gained while tracked · last 7 days
Pi 1.019.5×
+1,120
Gemini 4 Argon22.5×
+953
Sonnet 5.5432.0×
+750
Dots: Always-on agents132.5×
+497
+112
Also rose in the last 60 days
- Discovery of a new OpenAI agent message board+1,866
- AI-generated posters don’t have to be horrible+1,784
- GPT-6 Sol and Luna+1,531
- Introducing System One Models and Jev+1,331
- Claude Opus 5.5+1,294
- GPT-6 Astra+1,251
- Claude Fable 5.1 and Claude Mythos 5.1+1,249
- Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows+1,198
- GLM-5.3-Flash+1,019
- Claude, change the “Add to Cart” button to blue+940
- Anthropic's best AI model struggles to attract users as cheaper tools thrive+813
- Gemini 3.7 Flash+747
- How to Write with an LLM+746
- So you want to use OpenRouter?+715
- DeepSeek V4 Flash 0731+714
- Show HN: The load-bearing vocabulary of Claude+694
- Docker Sandboxes – Disposable, isolated sandboxes for AI agents+669
- Claude: System Prompts+656
- Cloudflare OS: an open platform for agents, apps, and work+613
- How I use LLMs to learn complex topics+580
- Your intellectual fly is open when you use an LLM to author a post (2025)+551
- I-have-ADHD: A skill to stop your coding agent from burying the answer+535
- Claude Code now reads AGENTS.md if there is no Claude.md+516
- Why your local LLM feels dumber than it is+505
- GLM-5.3 is now open-weight+498
- Cursor launches Origin, GitHub alternative+490
- GPT-5.6 Sol Pricing Cut by 50% on OpenRouter+484
- RAG Is Simpler Than You Think+476
- Muse: Meta's personal AI agent, features and capabilities+459
- Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots+453
- Claude is only available to people over 18 years+443
- Show HN: I trained a 125M model to autocomplete piano on-device+440
- Pion, an agent designed to run any company autonomously+435
- Tell HN: OpenAI keeps re-enabling the 'allow training' setting+378
- Porting my 1993 Amiga game to Godot, with an LLM reading the 68000 assembly+376
- Breaking Claude Code Opus 5 Auto Mode+355
- Desert Ant Labs: local, fast models that run on device+334
- Qwen 3.8 27B available on Cerebras at 1500 tokens/s+318
- fx :Tiny, open, native coding agent.+310
- Munder Difflin – Agent harness to run an office of your clones+309
- OpenAI Agents API+302
- Managing AI Coding Costs at Scale+298
- GPT-6 Astra on OpenRouter+293
- Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra+292
- Lost my phone at the office. Claude suggested tracking Bluetooth signal strength+289
- Claudette: Make Claude stop talking like a BuzzFeed article+281
- Maximizing the value of your Claude Code sessions+281
- Humans missed 1 in 3 threats approving AI agent commands across 40k game runs+272
- Which tools do Claude, Codex and Cursor choose? We measured 17k runs to find out+271
- OpenAI begins rolling out GPT-6 Astra+271
- GPT 5.6 Sol is the best "vision" model OpenAI ever released+259
- Claude Opus 5.5 Intelligence, Performance and Price Analysis (Max)+258
- Unreal Agent+245
- Learning more about Claude's mathematical capabilities+241
- Portal by Spotify cut my Claude Code token usage by 90%+234
- ChatGPT Work Tool and Skill Reference+234
- Learning Programming in an Age of LLMs+229
- Show HN: Huzzah – a novel approach to coding with AI+216
- Improving GPT‑5.6 Sol in ChatGPT, expanding GPT‑5.6 Luna access for free users+209
- Vomit: Clean up Claude 5's token output with a separate LLM+198
- Show HN: Running 104GB Qwen3.8-Flash-Next on 48GB Mac with at ~12 tok/s+189
- Show HN: We built open OpenRouter that turns usage into a better model+189
- Quick impressions: A week of using Codex more than Claude+185
- Best LLM for every budget, updated daily+181
- What I learned by putting GitHub Copilot behind a MitM proxy+171
- OpenAI: GPT 5.6 Sol price reduction (until at least Nov 21)+169
- Auto mode is now the default in Claude Code+164
- Anthropic appears to be A/B testing reduced effort levels in Claude Code+164
- Check if a file was made with Claude+159
- How well do agents use test/verification techniques?+142
- Models Are Getting Dumber on Purpose+139
- OCR It – pull text out of un-copyable documents for your LLM+137
- Claude Cowork and chat are now one Claude+132
- Show HN: ThoughtDAG – An editable context graph for LLM conversations+132
- OpenAI expands ChatGPT ads with Sponsored Agents+121
- OpenAI's GPT-6 Astra on ARC-AGI-3+117
- MathCode, Mathematical Coding Agent+117
- Mistral's Shieldstral: 3B open-weights model for multimodal moderation+116
- Claude Code weekly limits reduce by a third tomorrow+113
- Agent memory as a file format+113
- Apple Silicon and macOS VMs: Faster LLM Inference with llama.cpp+112
- Show HN: OwnTime – a chess clock for your day's priorities+104
- Yes, Claude can do nine loops+100
- Microsoft abandons personal AI chatbot race with Copilot reboot+99
- AI models ran real businesses: They sent $12,431 in fake invoices, lost $3,200+99
- Apple's Siri AI Can Be Swapped Out for Claude, ChatGPT, Code Shows+97
- GPT 5.6 Cyber+96
- Autolith: A programming agent with a live runtime+96
- Show HN: Ante, a coding agent in a single binary that runs offline+95
- Launch HN: Bullet (YC S26) – A Faster Coding Agent+95
- GPT-5.6 Luna vs. GPT-6 Astra: Is a $1.20 Model Good Enough for Code Review?+94
- Show HN: Engrim – A universal, local-first SQLite memory engine for AI CLIs+85
- Show HN: Nari Qwen3-TTS and Qwen3-ASR – High accuracy, low latency and cost+79