Mojo ๐ฅ 1.0 is in beta! Beta 1 marks the first step towards finalizing 1.0 later this year, which will bring a new level of language stability. The beta lands safe closures with a new capturing syntax, conditional trait conformance, and major variadic improvements. Plus, Mojo has its own home at https://t.co/j3TzPUkEDr.
Working through our GPU Puzzles? Don't sleep on our companion YouTube series that walks through puzzles 1 through 5. Follow along, pause, and rewind to make sure you grok the solution: https://t.co/2qQxNdhe6c
At @AMD AI DevDay, @clattner_llvm showed that AMD MI355X paired with Modular platform delivers equivalent image gen performance to Blackwell at 5.5x lower total cost.
Watch Chris' luminary talk: https://t.co/4oH8U0yAMr
Thanks again to @AIatAMD for a great event!
We were happy to sponsor #MLSys 2026. Across the talks, posters, and keynotes, three themes defined the current state of inference serving:
1. Agentic engineering
2. KV Cache optimization
3. Heterogenous hardware
Our read on each: https://t.co/GWRa8YKaFH
In the latest Modular Tech Talk, Mojo Compiler Engineer Billy Zhu presents Mojo's attribute-based expression system and how it enables Mojo's powerful type-safe meta-programming:
https://t.co/PyphPJZ8yv
One meeting: BLAS routines in Mojo, raylib Mojo bindings, consumer hardware support, MAX vs. vLLM, and where clauses in parametric structs. The community brought presentations, demos, and questions.
Catch up with the full recording of yesterday's community meeting: https://t.co/BdaRqLxvhi
Haven't had time to explore Mojo 1.0 beta yet? @InfoWorld published a piece on Mojo 1.0 that will get you up to speed on language basics, metaprogramming, Python interop, GPU support, and more:
https://t.co/NGtpTd4FHz
Tomorrow's community meeting is a deep dive into 3 community Mojo projects: a Mojo implementation of BLAS routines, Raylib v6 Mojo bindings, and replacing OpenCL with a Mojo kernel within Darktable, an open source photo editing program.
See what the community is building: https://t.co/N80NTSjokW
Seoul showed up! Packed room, sharp questions, a special message from @clattner_llvm, and an intro to Mojo ๐ฅ and MAX. Our first developer meetup in Korea. Thank you to SqueezeBits for making this happen, and to everyone who came out.
Traditional load balancers were built for stateless services, but LLM inference backends aren't stateless.
Part 2 of our inference routing series covers the data layer that queries live backend state across hundreds of pods at microsecond latency:
https://t.co/HBdAEf7XGH
Agentic engineering is changing what one developer can ship with Mojo in a few weeks.
@ehsanmok set out to build a pastebin service in pure Mojo ๐ฅ using our recently released first Mojo 1.0 beta. Using AI coding agents and Mojo's agent skills, he built 10 libraries from scratch: a full networking stack, SQLite bindings, high-performance JSON, reflection-driven serde, fuzz testing tools, and more. The app is live at https://t.co/hLDNqO1p4V.
Read about his stress test of Mojo 1.0 beta: https://t.co/hqpZEQkwga
AI agents in healthcare face tight constraints: latency can't exceed 800ms per turn, the first turn processes 10k tokens of context, and safety models analyze the conversation in parallel.
Using our MAX framework, @hippocraticai keeps patient conversations instant (sub-second TTFT), hits aggressive performance targets without sacrificing model accuracy, and runs across accelerators as new hardware comes to market.
A look at how regulated enterprises like Hippocratic AI use MAX in production for real-time patient conversations:
https://t.co/xllxNL3MTK
The MAX-LLM book just made it even easier to build an LLM from scratch.
The new notebook format lets you run the GPT-2 components interactively, inspect real tensor shapes, and generate text from pretrained weights.
Prefer to browse first? The pre-rendered version shows all outputs without running a cell:
https://t.co/V3tysngguO
๐ Ring-2.6-1T is now open source.
A trillion-scale flagship thinking model built for real-world complex tasks: Agent workflows, coding & engineering, long-horizon tasks, complex reasoning, research, and enterprise automation.
It is designed to move beyond โansweringโ toward execution: understanding context, planning steps, calling tools, and staying stable across long task chains.
Highlights๏ผ
- Advanced agentic workflow support.
- Reasoning effort levels: high for agentic tasks, xhigh for complex reasoning.
- Scalable asynchronous RL via the IcePop algorithm, enabling stable, trillion-scale training for long-horizon agentic RL.
Two can't-miss events coming up in Seoul:
๐ Modular's Judy Heflin presents an intro to Mojo & MAX at the Efficient AI Offline Meetup on May 16th at Dream Plus Gangnam: https://t.co/w9lpuL3EJq
๐ Our inaugural Modular Developer Meetup in Seoul on May 19th at Belgium Jazz Cafe: https://t.co/5ztBwpJeFv
Come for the talks from Modular & SqueezeBits. Stay for the NVIDIA RTX 5080 + AMD Radeon RX 9070 raffles ๐
Mojo has minimal boilerplate, a strict type system, and compile-time validation of code, all things that make it well-suited for use with AI coding agents.
We're taking this up a level by publishing a set of Mojo agent skills that make translating code to Mojo a breeze.
Full writeup + CUDA kernel โก๏ธ Mojo translation demo: https://t.co/fKERenlRXH
What would you build with lightning fast image generation?
Inkwell is @iamtimdavis' answer: a dynamic storybook-building app that uses @bfl_ml's FLUX2 and @googlegemma 4 to write and illustrate in real time. Powered by Modular Cloud.
We sat down with Tim to talk through how Inkwell works under the hood: https://t.co/iQOZSLu7QS
The short version: LLM tokens stream directly into the image prompt before the story finishes generating. First pixel under 500ms. Built on Mojo kernels and MAX serving infra.
Bolt's big dark: your best friend Bolt is a small silver robot with one wobbly antenna and a tiny light on his chest that blinks when he's nervous. He says the dark feels too big and too quiet and he doesn't know what's in it. Bedtime is in 10 minutes, and it's up to you to reassure him that everything will be okay:
https://t.co/Sc4OOwLMvJ