Other softwareAI models and assistantsLibraries and frameworksOperating systems and desktopMedia and creativeAI coding toolsInfrastructure and DevOpsTerminal and shellBrowsers and extensionsEmail and chatSelf-hosted and home serverNetworking and VPNEditors and IDEsSecurity and passwordsProductivityDatabasesFiles, sync and backupGit and code hostingNotes and knowledgeMonitoring and analyticsSearchFinance
AI models and assistants
What HN commenters recommend, ranked by distinct recommenders. Under each: what people praise most and the most common caveat, or the strongest quote where there is no summary yet.
- 1.Loved for Agent orchestration, Ambitious coding · Caveat Condescending tone
- 2.Loved for Planning and orchestration, Code quality · Caveat Cost and limits
- 3.Loved for Non-coding tasks, Search replacement · Caveat Coding and agents
- 4.Loved for Low cost, Good-enough quality · Caveat Hard tasks
- 5.Loved for Frontier alternative, Code review · Caveat Claude-like prose
- 6.Loved for Coding help, Search and research · Caveat Errors, hallucinations
- 7.Loved for Low cost, Speed · Caveat Below frontier
- 8.Loved for Coding ability, Cost and limits · Caveat Stops early
- 9.Loved for Cost value, Code review · Caveat Hallucinations
- 10.Loved for Daily driver, Consumer hardware · Caveat Below frontier
- 11.Loved for Low cost, Coding quality · Caveat Weaker than Opus
- 12.Loved for Frontier-level quality, Coding daily driver · Caveat Cost and tokens
- 13.Loved for Ollama alternative, Performance tuning · Caveat GPU setup
- 14.Loved for Cost efficiency, Routine workhorse · Caveat Limited intelligence
- 15.Loved for Runs locally, Coding ability · Caveat Overthinking
- 16.Loved for Speed, Coding · Caveat Usage limits
- 17.Loved for Cheap metered usage, Easy model switching · Caveat Unreliable routing
- 18.Loved for Local hardware, Writing quality · Caveat Compaction failures
- 19.Loved for Speed, Price · Caveat Company trust
- 20.Loved for Coding quality, Planning and review · Caveat Usage limits
- 21.Loved for Free and open source, Parakeet support · Caveat Accuracy limits
- 22.Loved for Personal assistant, Chat app control · Caveat Security
- 23.Loved for Local hardware, Quality for size · Caveat Agentic coding
- 24.Loved for Cloud plan value, Local models · Caveat Attribution concerns
- 25.Loved for Personal assistant, Better than OpenClaw · Caveat Model-dependent quality
- 26.Loved for Low cost, Daily driver · Caveat Slow speed
- 27.Loved for Knowledge and chat, Refactoring and review · Caveat Quality degradation
- 28.Loved for Low cost, Speed · Caveat Complex tasks
- 29.Loved for Subagent workhorse, Cheap and fast · Caveat Complex coding
- 30.Loved for Consumer hardware, Small-task models · Caveat Looping
- 31.Loved for Speed, Runs locally · Caveat Accuracy vs Whisper
- 32.Loved for Local API server, Easy setup · Caveat Closed source
- 33.Loved for Low cost, Coding capability · Caveat Slow, overthinks
- 34.Loved for Code review, Core logic · Caveat Verbose output
- 35.Loved for Low cost, Speed · Caveat Average accuracy
- 36.Loved for Low cost, Agentic tasks · Caveat Coding quality
- 37.Loved for Runs locally, Accuracy · Caveat Transcription errors
- 38.Loved for Home visualization, Image cleanup · Caveat Spatial prompts
- 39.Loved for Grounded document Q&A, Audio overviews · Caveat Flaky UX
- 40.Loved for Natural voice quality, Audiobooks and articles · Caveat Pronunciation errors
- 41.Loved for Low cost, European alternative · Caveat Behind frontier
- 42.Loved for Runs locally, Speed · Caveat 20B mistakes
- 43.Loved for Fast prototyping, Claude Code pairing · Caveat Generic output
- 44.Loved for Large local models, Daily reliability · Caveat Generation speed
- 45.Loved for Coding ability, Price · Caveat Data training
- 46.Loved for Local and private, Good results · Caveat Below frontier
- 47.Loved for Everyday coding, Multi-model workflows · Caveat Hard tasks
- 48.Loved for Quick answers, Replaces search · Caveat Unreliable results
- 49.Loved for Tool integrations, Removes drudgery · Caveat Bugs
- 50.Loved for Search grounding, Model choice · Caveat Limited scope
- 51.Loved for Errand automation, Ongoing monitoring · Caveat Privacy
- 52.Loved for Batched throughput, Production use · Caveat Setup effort
- 53.Loved for Mac local runtime, Ease of use · Caveat Qwen tool calls
- 54.Loved for Speed, Coding assistance · Caveat Rate limits
- 55.Loved for Speed, Price-performance · Caveat Deprecation
- 56.Loved for Simpler coding help, Everyday writing tasks · Caveat Unreliable coding
- 57.Loved for Quant quality, Fixes issues · Caveat Context room
- 58.Loved for Speed, Low cost · Caveat New model lag
- 59.Loved for Low false positives, Overall quality · Caveat Misses AI text
- 60.Loved for Search replacement, Source citations · Caveat Declining quality
- 61.Loved for Agentic tool use, Model quality · Caveat Speed and hardware
- 62.Loved for Consumer hardware, Output quality · Caveat Vision accuracy
- 63.Loved for Catches issues, PR integration · Caveat Missed bugs
- 64.Loved for MCP integrations, Non-engineer friendly · Caveat Bugs
- 65.Loved for llama.cpp companion, Dynamic model swapping · Caveat Lemonade alternative
- 66.Loved for Runs locally, Small model options · Caveat Slower locally
- 67.Loved for Agent orchestration, Control and reliability · Caveat Reliability effort
- 68.Loved for Broad ecosystem, Practical integrations · Caveat Unnecessary abstraction
- 69.Loved for Simple tasks, Personal data · Caveat Fails basics
- 70.Loved for Voiceovers, Voice quality · Caveat Real-time limits
- 71.Loved for Pluggable providers, Local models · Caveat Needs tinkering
- 72.Loved for Transcription workflows, Strong recommendations · Caveat Bugginess
- 73.Loved for Personal songs, Cover feature · Caveat Hollow output
- 74.Loved for Creative collaboration, API integration · Caveat Model removal
- 75.Loved for Codebase understanding, Flow diagrams · Caveat Odd diagrams
- 76.Loved for Fast to working app, Non-coder friendly · Caveat Competition from agents
- 77.Loved for Extensible, Works anywhere · Caveat Not agentic
- 78.Loved for Speed, Model choice · Caveat Policy changes
- 79.Loved for Privacy, Free access · Caveat Weaker models
- 80.Loved for Transcription accuracy, Daily workflow · Caveat Declining accuracy
- 81.Loved for Production workloads, Good value · Caveat Model deprecation
- 82.Loved for AMD/Strix Halo support, Many backends · Caveat Small default context
- 83.Loved for Strong local model, Agentic coding · Caveat Not frontier-level
- 84.Loved for Daily driver, Low cost · Caveat Not top-tier
- 85.Loved for Image quality, Complex edits · Caveat Face distortion
- 86.Loved for Low cost, Privacy terms · Caveat Model coverage
- 87.Loved for Output quality, Token efficiency · Caveat Slow generation
- 88.Loved for Local models, Accuracy · Caveat Quirks
- 89.Loved for Design to code, Agent integration · Caveat Manual finishing
- 90.Loved for Security research, Thorough, goal-focused · Caveat Usage limits
- 91.Loved for Coding implementation, Speed · Caveat Needs guardrails
- 92.Loved for Catches misses, Hard problems · Caveat High cost
- 93.Loved for Fast transcription, Talking to agents · Caveat macOS only
- 94.Loved for Free access, App prototyping · Caveat Confusing UX
- 95.Loved for Unified LLM proxy, Usage monitoring · Caveat Supply-chain security
- 96.Loved for Coding help, Everyday troubleshooting · Caveat Forced integration
- 97.i use this for my personal projects. some features are gated behind a license but the basics like provider proxy, logs, metrics are covered in the free version.
- 98.I have been using Sites a few months now, really underrated feature IMO.
- 99.I'd rather just use cheap tokens from OpenCode Zen and Go.
- 100.Thanks to faster-whisper and local models using quantization, I use it in all places
- 101.It one-shotted generation of Java bindings for this project
- 102.I use Spokenly, offline-only mode with the Nvidia model. All local, totally free. Highly recommend
- 103.now I use a custom consult document picker (with rich annotations) to perform actions on different kinds of XML documents
- 104.I've spent 374M tokens this month and it only cost me $18 on energy-based pricing.
- 105.Been using Aegis for a while now, the encrypted backup alone makes it worth it over Google Authenticator.
- 106.I use the humanize skill to clean up AI written work before handing it over to colleagues.
- 107.which is just 100M params, and excellent in both speed and quality (English only). I use it in my voice plugin
- 108.I use the official Blender MCP to drive Blender from Claude.
- 109.my sleep quality got much better when I turned on f.lux.
- 110.I used to be a big advocate for Salesforce in my organization. And it was really great
- 111.I like “talk to it while I’m driving” format a lot
- 112.I use Google Assistant for things like "add a reminder", "set an alarm"
- 113.for all day to day queries and prompts I have switched to using Proton's Lumo that is really quite good
- 114.I have used NanoClaw running on my Mac to do the following:
- 115.Mistral OCR is really good and their small models are great for simple translations, I use them regularly.
- 116.Please do yourself a favour and check out GLM-5.2, Qwen-3.7-Max, MiMo-2.5-Pro, and DeepSeek-V4-Pro.
- 117.https://synthetic.new with Kimi K2.5 works surprisingly well for small tasks where I still make all the decisions.
- 118.I love the idea of a gaming box that can host multiple instances; that would be very very good.
- 119.you'll realize this is obviously the future interface.
- 120.QWEN3 TTS has a “clone” feature. You can use the local version, give it a voice reference and text
- 121.Check out the awesome [Claudish-to-English plugin]
- 122.I've been using https://github.com/mastra-ai/mastra which is pretty similar but has workflow visibility and a number of templates.
- 123.I use gpt image 2 very heavily for my current project (ai UI design tool).
- 124.i use qwen-30b-vl all the time locally as my goto model because it's just so insanely fast
- 125.Unsloth Studio [0] is what I recommend these days, open source alternative to the more widely known LM Studio
- 126.I expect I'd still like playing with Midjourney for more artistic styled work but it was the cost that drove me away.
- 127.I love cloudflare ai gateway, been using it to bring some ai into the functionalities.
- 128.It works very well, the integration is much easier than before, users have model choice.
- 129.llama-server also comes with a truly excellent built-in web chat interface these days
- 130.Mistral has been great for us as a provider for our EU-based company. We use Mistral Large 4, GLM 5.3 Flash and DeepSeek v4.1 Flash 1017 through their API
- 131.Strongly recommend https://github.com/Neroued/ninfer, which can pull ~180 TPS on 5090 with 3.8
- 132.for my needs, nothing could beat Whisper Large v3.
- 133.On macOS I've been using piper (https://github.com/OHF-Voice/piper1-gpl) to announce claude code notifications and it works perfectly!
- 134.Check out chatjimmy.ai blow your mind. I’m not affiliated.
- 135.It was the best product in its area from the beginning, eclipsing anything US produced prior to it
- 136.You run OpenCode with Big Pickle on it with decent performance.
- 137.I'd also recommend checking out KittenTTS[2], I use it and it's great for the size/performance.
- 138.I do an insane amount of work with my little $20/mo. ChatGPT Plus subscription and never hit limits.
- 139.It's not perfect as I have to manually do the correction like always. However, it definitely saves me some time and effort.
- 140.One week I created 4 different scripts with Gemini Canvas to automate some tedious work, it took my I don't know, 3 hours?
- 141.I really like https://ai.pydantic.dev/ as it builds on existing developer knowledge and there is not much to learn in the framework itself.
- 142.Step 3.7 Flash at IQ4_XS beats Qwen 3.6 27B despite the latter having MTP, on all of my actual coding task evaluations in real codebases.
- 143.With FIBO2 and Qwen Image 2.1, I can transform my raw shots so they look as though they were captured with a high-end digital camera.
- 144.Olmo and to a lesser extent Nemotron are what you should look at
- 145.I've actually seen better performance with older ones like Meta Llama 4 or Amazon's Nova
- 146.Very nice, multi modal, largest open weight model that supports audio.
- 147.If you want something different that is designed for levels, check out Marble by World Labs.
- 148.I’ve really appreciated its qualities when supervising my agents’ activities
- 149.Olmo and to a lesser extent Nemotron are what you should look at
- 150.a better known implementation (in Bash) is probably https://github.com/holman/spark
- 151.No other model, open or closed, has been able to hit such a low AER (Acronym Error Rate ;)) for my meeting transcripts.
- 152.Checkout MTPLX. I use the Youssofal/Qwen3.8-27B-MTPLX-Optimized-Speed model and I get around 12 tok/s on a 36GB MBP.
- 153.I run a ClawdBot (Claude on a Mac Mini) as a 24/7 AI Chief of Staff for my company.
- 154.Bonsai makes the vdom incremental and it is pretty fun to work with.
- 155.I use Gemma 4 12B in the 4-bit quantization for all sorts of vision tasks (image sorting, classification, description).
- 156.Python has won for me: fastapi, pytorch, langchain, streamlit, and so on and on.
- 157.I think the plain ordinary chatbot behind the Copilot on the desktop is fine