How I build apps these days: 1. Record a 1-5 minute prompt with @WisprFlow The more detail, the better. 2. Add visual inspo so the agent knows what I have in mind (screenshots or Inspo MCP). 3. Use @mattpocockuk's grill-me skill to have the agent ask clarifying questions. 4. Get back a v1, then send a list of feedback, feature requests, and design tweaks (via another voice recording usually). 5. For each new feature, ask for a quick screen recording to verify it works before merging the PR. 6. Keep iterating until I'm happy, then post it on @X or throw it away. For my harness, I use a mix of Claude Code with Opus 5.5 (high) and OpenCode with DeepSeek V4.1 Flash / GLM 5.3. I also give my agents some CLIs & skills: • Vercel + GitHub CLIs so they can push code and deploy apps. • Together skills so they can use chat, image, & voice models from Together • Design/animation skills so they have guidance on spacing & interactions

Very excited to announce the new post training platform from @togethercompute. Together's team has deep experience in training, and have done seminal work in the field, from building the world's largest open datasets like RedPajama, new new architectures like Mamba, that have made their way into the world's leading models, to innovations in training kernels like FlashAttention. Our new training platform draws on that experience and talent to offer the most flexible and highest quality post training platform on the market today. Together Post Training, combined with our training FDE team, has been helping companies own their AI models and we'd love to help you if you are considering moving from rented to owned intelligence.

This model is insane at design. And 88x cheaper than Claude Opus 5.5 on my test! I asked M3.1 Flash (left) and Opus 5.5 (right) to build me a landing page for a hardware device launch. MiniMax cost just $0.014 vs $1.22 for Opus, while being 4x faster and more token efficient.

Since this went a bit viral, here's a 1 minute demo! Launching soon. 100% free & open source. Gonna send out some early access invites this weekend. Lmk if you wanna try it out & send feedback!

Excited to share that I'll be speaking at @aiDotEngineer NYC next week... with my dad! We'll be giving a joint talk on using open models to forecast a company's financials. See you in NYC!

Hit 1,000 days at @togethercompute last week! 🎉 When I joined, I was still writing code by hand. Crazy how much has changed in such a short time. Grateful to be building the future of AI with an incredible team!

Introducing GeoGuess Bench! A benchmark to measure how good AI models are at playing GeoGuessr. We gave each model 210 photos from around the world, asked it to predict where each photo was taken, and scored it based on how close it got. Some surprising results: • Muse Glimmer 30B beat GPT 6 Astra while being ~45x cheaper • GLM 5.3 Flash matched GPT 6.1 Sol while being ~5x cheaper • Claude Opus 5.5 took the #1# spot, scoring even higher than Fable 5.1 Open models did surprisingly well overall. Check out the full leaderboard, Pareto frontier curve, and every model's guesses! - Site → - Code →

We've been quietly rebuilding hallmark. Since launching a few months ago, it's been installed 50,000+ times! @YoussefUiUx & I have spent the last few months listening to feedback & building v2. Hallmark v2 is dropping soon.

Just open sourced Tev1 0.8B! Weights are below. Feel free to download them. Definitely not as good as Tev1 4B or Jev, but it runs entirely locally on a mac. It can work decently well for very simple classification.

Opus 5.5 is incredible at launch videos. I gave it an app I launched recently & it one-shotted a great launch video for me. Was very impressed with the animations & charts. I love open models and use them a lot, but I'm also a big believer in using the right model for the job

Just trained Tev1 0.8B, a tiny Jev-like classifier. Here it is running completely locally on my mac with @ollama & classifying some tasks. It's extremely fast: only ~50ms E2E latency. Video is not sped up! Releasing weights & benchmarks very soon so you can try it yourself :)

We're releasing tev1-4B-experimental, a Jev-like classifier finetuned on top of Qwen3.5 4B. We're making it available on Together serverless at $0.042/M input & $0/M output. Also releasing the data recipe & a tutorial on how to finetune your own (Tev1 cost $17 to train!).