AI Models

"I don't use this for photorealistic stock footage, but it's fantastic for bringing stylized or artistic images to life. The June 2026 update makes animating those stylized images remarkably easy."

"Grok's image engine remains a hit-or-miss tool, but it deserves an honorable mention. If you need physics-accurate photorealism, go elsewhere. But for highly creative or abstract motifs, Grok often surprises you with the most interesting result."

"I've been running Sonnet 5 against Opus 4.8 on real tasks this week, and it's closer than any Sonnet generation has been before. It's become agentic in a way earlier Sonnets weren't — it strings together more tool calls, sticks with harder problems longer, and rarely needs me to step in and redirect it. For most Claude work, this is the model to reach for by default."

"Opus used to be my clear go-to for complex architectural work, but between Anthropic's newer tiers and how good Sonnet 5 has gotten, it's been squeezed into a narrower niche. Still incredible at writing and still worth reaching for on the hardest problems."

"Haiku isn't meant for writing poetry or solving hard math. It's meant to be a rapid-fire processor. I use it to parse messy text, summarize long documents quickly, and handle routine categorization."

"Working with Fable is a genuine pleasure. It has a different 'personality' than other models — almost mind-reading when it comes to figuring out what I want — and it holds a whole complex problem without dropping a detail. It was pulled by a US government order for a few weeks in June, but it's back now, and I save it for high-value work: audits, planning, and deep concept work."

"It's almost suspiciously cheap. I use it through OpenRouter when I need to run a task hundreds of times and don't want to burn through API credits. The reasoning capability at this price point is completely disruptive."

"ElevenLabs is still the undisputed king of text-to-speech. The v3 model finally added proper inline audio tags, meaning I can just type [laughs] or [whispers] and it executes it flawlessly. It feels like directing an actor."

"When I need a photo that genuinely looks like it was shot on a DSLR and not generated by an AI, FLUX-2-Pro is still one of my top choices. It gets the lighting and physical textures right in a way other models struggle with."

"Fugu isn't a traditional model — it's somewhere between a tool and a model. It acts as a 'conductor,' deciding which model (Claude, GPT, or an open-weights model) is best suited for your prompt and routing it there dynamically. I choose it as the engine to power my agents when I want the absolute best result without worrying about which API to call."

"A fantastic open-weights option for coding. I've been experimenting with it via OpenRouter, and it handles long-horizon tasks surprisingly well. It's a testament to how fast the open-source community is moving."

"GPT-5.5 is the reliable workhorse. I don't use it for my most complex architectural planning or creative writing, but if I just need something formatted, summarized, or translated quickly, it rarely misses."

"This is currently my primary tool for image editing. If I need a specific object removed or a color changed, I can just tell it what to do in plain English rather than messing with masks."

"I use 3.5 Flash for some of my work. It's great for analyzing my UX/UI designs and for writing consistent copy across multiple languages. It's blazing fast, though I switch to stronger model for the truly difficult logic problems."

"3.1 Pro is my choice for challenging work where I need a second or third opinion. Its massive context window means I can just dump an entire codebase or project folder into it and ask high-level architectural questions."

"When I need a bit more brainpower than the 12B can offer, I spin up the 27B variant. It's noticeably smarter for logic tasks, though you feel the hardware requirements a bit more."

"I run this locally on my computer. It's smart enough for most admin and copy work, and the fact that it's multimodal means I can use it for local automation tasks without sending any data to the cloud."

"The structural control is what makes Lyria 3 Pro stand out. Being able to define exactly when the chorus hits or how the bridge builds up makes it a genuine tool rather than a toy."

"Midjourney is still the undisputed king of 'pretty.' I don't use it for functional UI mockups anymore, but if I need a hero image for a blog post or concept art that looks like a painting, nothing else comes close."

"Nano Banana 2 changes the game entirely. The fact that you can ask it to dump out ready-to-use screen mockups or detailed infographics makes it arguably the most useful visual tool available today for a professional workflow."

"A solid middle ground. If I need something to run a basic logical check or parse some structured data reliably without paying the premium o3 tax, this does the job nicely."

"When I hit a wall with a complex coding problem or a logical puzzle that other models fail on, o3 is the heavy artillery I bring in. It's too slow and expensive for daily chat, but invaluable for the hard stuff."

"I use Qwen3.7-Max via the cloud when I need an agent to really dig into a codebase. The smaller distilled versions are also great for running locally via HuggingFace when I need to keep things offline."

"When I need answers grounded in real-time facts rather than model pre-training, Sonar Pro is my default. It powers Perplexity's Pro search, and its true superpower is how it translates natural language questions into optimized search queries and then cleanly summarizes the findings with proper citations."

"When I need a short video clip for a presentation that genuinely looks like a stock video site, I use Veo. The physics coherence is incredible."