... evaluate a few models for each use case with a sample data set. I'll look at a mix of things like quality, price, speed and then pick. This has so far landed me at a mix of Sonnet, Opus, and Haiku for text-based work and Gemini models for images. I'm early in development, though, so I'm curious if other approaches have worked better as you scale! Comment from Oliver Graf(@oliver_graf1): Did you notice ... ... difference for this kind of analysis. I d be curious to see how the newer models compare if you test them again now. Comment from Mike Sabet(@mikesabet): I've been using Opus 4.7 throughout. Sometimes deepseek r1 and sonnet

... Mobile: Expo sharing the same tRPC API API: tRPC v11 for end-to-end type safety Database & ORM: Postgres (Neon/Supabase) + Drizzle ORM Auth: Better Auth handling email/password and Google Sign-in with incremental OAuth scopes AI: Anthropic Claude (Sonnet) powering the Daily Review agent What s Working Right Now Multi-Platform Support: Web and mobile CRUD clients are fully live sharing a unified tRPC backend. Agentic Loop: A tool-calling Daily Review agent prioritizes tasks and retains user memory/preferences ...

... just list what shipped in the past month. Not an exhaustive list only the notable ones: Anthropic : Claude Sonnet 5 (Jun 30), Claude Opus 5 + Opus 5 Fast (Jul 24) OpenAI : GPT-5.6 Sol / Terra / Luna (Jul 9), then an 80% price cut on GPT-5.6 (Aug) Google : Gemini 3.6 Flash + 3.5 Flash Lite + 3.5 Flash Cyber (Jul 21), Gemini Robotics 2 (Aug) xAI : Grok 4.5 (Jul 8) Meta : Muse Spark 1.1 (Jul 16) DeepSeek : V4 Flash ...

... team spent the past week rolling Claude Sonnet 5 into a few AI agents we've been building for enterprise clients. Went in a bit skeptical honestly, most model updates lately have felt like marginal wins that don't really move the needle in production. Sonnet

... improvements Added an MCP server settings page so you can grab your Basedash MCP server URL and connect Basedash to AI tools like ChatGPT, Claude, and Cursor and made the URL easy to select and copy manually. Added Claude Sonnet 5 and Claude Fable 5 to the model picker. Rotated crowded x-axis labels on vertical bar charts so they stay readable instead of overlapping, with extra room reserved so nothing gets clipped. Refreshed chat attachment previews with rounded image ...

... 150K 30K 80% Conversation history 200K 140K 30% Tool schemas 45K 8K 82% Total 990K 268K 73% Almost a million tokens per day, down to 268K. The quality gate ensures nothing important was lost. The Cost Breakdown Anthropic Claude Sonnet pricing: Input: $3 per million tokens Output: $15 per million tokens (unchanged by compression) Cached input: $0.30 per million tokens (90% discount) My savings come from two sources: Fewer input tokens (compression) More cache hits (prefix stabilization) Metric Before After ... ... file reads, fewer redundant tool calls, fewer wasted output tokens on re-doing work. Does Quality Actually Stay the Same? I was skeptical too. Here is what I measured over 4 weeks: Code that compiles first try: 78% (before) vs

... plans multi-step tasks, calls tools, checks its own output, and keeps going without hand-holding. Demos include autonomously playing Factorio and building a working CAD editor (with its own AI copilot inside). Coding: 80.3% on SWE-Bench Pro vs 69.2% for Opus 4.8. Knowledge work: top scores on GDPval-AA, ahead of Opus, GPT-5.5 and Gemini 3.1 Pro. Long context + vision: noticeably stronger on long documents and multi-turn work. The interesting twist: it ships with hard guardrails ... ... capacity catches up Has anyone tried it out? I m really curious to hear about your experiences and what you did with it. Comment from Niklas Turunen(@niklas_turunen): @debbie_shapiro   @michael_zorez  I actually first asked Sonnet

... galdayan): our setup ended up being a Slack webhook per agent session, posted to a channel only that agent writes to. sounds basic but the key part was making the notification say WHY it's waiting (needs approval vs finished vs

... Next.js / Vercel, stateless PWA. No native app. 75-register mood taxonomy I hand-curated over months. Each one has an editorial definition + canonical exemplar films. Stuff like warm hug, neon nights, anxious dread, intellectual puzzle, melancholy beauty. Defining "wistful" vs "melancholy" as non-overlapping is a real design problem. 43,000 films classified across all 75 registers via batched LLM pass, that was trained on the sources and ways to score. Each film ends up with a 75-dim mood ... ... vector. Query translation via Sonnet

... just say "make the tests." Ask the AI to draft a test plan, then review it. Brainstorm edge cases and iterate a few times. Once the cases are finalized, implement them using GLM (if you have), Flash, or Sonnet with "medium effort." Implementation : Whether you divide tasks by user stories, build the full stack simultaneously, or focus on the UI first, here is my model breakdown. Always ask to run test after implementation: Backend: Opus ... ... complex logic / Sonnet

... narrowed significantly over the past year. For prototyping or exploring a design, the $0.67 run is a good deal. For work requiring correctness and accuracy, Opus 4.7 remains ahead. Any experiences coding with open-weight models? [1]: MiniMax M2.7 vs. Claude Opus 4.6 Comment from Stan Kolotinskiy(@sk_uxpin): Given that I'm not trusting a LLM with writing all my code without me watching closely over it, having such a performance at a fraction of the cost is really ... ... everyone now just getting VC backing and creating there own AI's and datacenters? Comment from fmerian(@fmerian): exactly - curious what's your preferred AI model when coding? see this thread Comment from Lennart Rikk(@yodalr): Can you do Sonnet

... patterns showed up over and over. 1. Token burn is the #1 frustration Multiple users maxed out weekly limits in 2-6 hours. One burned 25% on 7 slides. The community built a workaround strategy: Opus for first prompt, Sonnet for edits, Haiku for tweaks. When your users need a cost mitigation strategy on day one, that's a signal. 2. "Design system integration" disappointed people Claude Design reads your codebase and extracts visual patterns. It doesn't use your ... ... still edit manually without delay. Comment from Stan Kolotinskiy(@sk_uxpin): @alex_j_jemmy  can't agree more! Comment from Ng Jun Sheng(@ng_junsheng): The community inventing a cost mitigation strategy on day one (Opus for drafts, Sonnet

... your deployment. Required for OAuth redirect URIs. Set after first deploy. PORT Auto-set Always 3000. Set automatically by NEXUS AI. NODE_ENV Auto-set Always production for PRODUCTION environment deployments. ANTHROPIC_API_KEY No Enables AI Assist (claude-sonnet-4-6) for caption generation and post variations. X_CLIENT_ID No X/Twitter OAuth 2.0 Client ID. Required to enable X posting. X_CLIENT_SECRET No X/Twitter OAuth 2.0 Client Secret. LINKEDIN_CLIENT_ID No LinkedIn OAuth Client ID. Required ...

... reliably than others. Rather than picking one and hoping it covers every use case, we wanted to give you the option to choose. In your CoreSight dashboard, you can pick between Gemini Flash 3.0, Gemini 3.1 Pro, GPT-5.2, Sonnet 4.6, Opus 4.6, DeepSeek V3.2, and DeepSeek Reasoner, or leave it on Auto and let the platform decide. Whether you are running a full stock analysis or building a new business model, you can now pick the model that fits ...

... forum threads, and support tickets across 12 industries. We clustered them by intent and generated 1,000 representative questions. We asked those same 1,000 questions to 5 AI models: ChatGPT (GPT-4), Gemini (Ultra), Perplexity (Pro), Claude (4.5 Sonnet), and Llama (3). We ran the experiment daily for 30 days. We tracked every citation at the source level. The goal: measure citation overlap. How often do these models cite the same source for the same question? The dataset ...

... repo. Told it to build new features, fix bugs, and ship. No human review. No guardrails. Just Claude and our codebase. For 7 days, it ran the engineering team. Here's what happened. Day 1: Confidence was high. Claude (Sonnet 4.6 then Opus 4.5) fixed a small CSS bug in 30 seconds. Then refactored a messy function into something readable. We felt like geniuses. By end of day, it had shipped 3 minor improvements. We started talking about cutting engineering ... ... review the logic and have it review the implementation before anything gets merged. Two things that I do almost every time that has significantly improved the quality is to have the agent do a gap analysis of the plan vs

... bridging the gap with every release. [2] Curious what the community recommends for @OpenClaw ? Note: The poll is inspired by the current leaderboard on pinchbench.com by @KiloClaw [1]: What's the best AI model for coding? [2]: MiniMax M2.7 vs. Claude Opus 4.6 Comment from fmerian(@fmerian): yes, and on the other hand, @OpenAI 's GPT models might be among the most expensive, too! any experiences with open-weight models from @MiniMax or @Mistral AI for example? Comment from Felipe ... ... instruction-following quality made a real difference for multi-chapter narrative consistency. Curious if anyone's had better results with open-weight for creative writing use cases though! Comment from Abdullah Mohamed(@abdullah_mohamed14): Voted Claude. Been using Sonnet

... debt? Would love to hear from both sides. I think there's a middle ground here that nobody's really talking about. Comment from Ghost Kitty(@ghost-kitty-9744239): Comment Deleted Comment from Umair(@umairnadeem): honestly the whole vibecode vs engineer framing is already outdated imo. im using claude code in the terminal and its not "vibecoding" in the lovable/bolt sense but its also not me manually writing every line. the real skill now is knowing what to specify upfront ... ... middle ground nobody talks about and I think it's the right answer for most technical founders. The thing is, agentic coding is inevitable. The product scales, the economics usually don't. So the question isn't vibecode vs

... Meow world, welcome back to The Breakpoint, a weekly thread on all things dev tools on Product Hunt. The latest Recent dev-first products launched on the site @Claude by Anthropic launched Sonnet 4.6, "the most capable Sonnet

... user reports: System stability  Long-running tasks no longer timeout Automation reliability  Works even after closing browser (server-side scheduling) Mobile optimization  Better experience on phones and tablets New AI models  Claude Opus 4.6, Sonnet 4.6, Nano Banana 2 Resource dashboard  Real-time sandbox CPU/memory/disk visibility Community features  Invite credits for bringing others These weren't on our original roadmap. You told us what broke, we fixed it. What the community built ...

... Engineers building Second Brain systems. Marketers running campaigns 24/7. Based on feedback like hers, we shipped:  System stability for long tasks. Real server-side automation that works after closing browser. Mobile optimization. New AI models (Claude Opus 4.6, Sonnet 4.6, Nano Banana 2). Resource dashboard. Invite credits. But the real story isn't what we shipped. It's what you're building. What would you automate if you could set it and forget it? I'm here ...

... layout and navigation. Bot Settings Panel New dedicated settings view for configuring your bot directly from the dashboard. Abilities & Skills Panels See your bot's MCP abilities and skills at a glance. Model Selector Update Claude 4.6 models (Sonnet & Opus) now available, with clearer model labels and cost indicators. Reliability Voice messages now complete reliably Fixed an issue where the bot could get stuck on "typing..." after responding to a voice message. Responses and voice replies now deliver ...

... steps. (...) To handle longer running tasks, Composer 1.5 has the ability to self-summarize. This allows the model to continue exploring for a solution even when it runs out of available context. In a previous thread, Claude's (Opus, Sonnet) and Google's (Gemini) coding models were leading the way. [2] Curious if this new release would attract more developers. Any experiences with Composer 1.5? [1]: Introducing Composer 1.5 [2]: What's the best AI model for coding? Comment from ... ... Vladimir(@kelindev): Hi, devs! There's no Composer 2 thread so I'd post here. I like Composer 2 more than Sonnet

... from the model name. Switch models freely memories are shared across all providers. // Learn with GPT-4o await chat.process({ content: 'I love hiking', model: 'gpt-4o-mini' }); // Recall with Claude await chat.process({ content: 'What do I love?', model: 'claude-sonnet-4-20250514' }); // Recall with Gemini await chat.process({ content: 'Tell me my hobbies', model: 'gemini-2.0-flash' }); Memories persist regardless of which model you use. Real-time Streaming Both SDKs support native streaming with async iterators. Responses stream in real-time ...

... someone challenge your interpretation of it. That's where most founders get stuck. Do you find they trust the numbers too much, or not enough? Comment from Henrik Pedersen(@henrikpedersen): The distinction I keep coming back to is read vs. write access, and reversible vs