Olmo 3 is a fully open LLM Olmo is the LLM series from Ai2—the Allen institute for AI. Unlike most open weight models these are notable for including the full training data, training process and checkpoints along …

Olmo 3 is notable as a "fully open" LLM - all of the training data is published, plus complete details on how the training process was run. I tried out the 32B thinking model and the 7B instruct models, + thoughts on why transparent training data is so important simonwillison.net/2025/Nov/22/...

Audience capture is really powerful and no one is immune to it.

Have a "posses" feature where you get to take over a pawn (unsure what for other than experiencing your colony as a colonist)

I want to like dspy but I'm thrown off by the obfuscated terminology and the API trying too hard to be Pytorch. It makes me feel that it is trying to pass as something more complicated than what it is. I wouldn't even call it optimization, it's closer to an autotuner from what I gather.

September 2025 dev update godot-rust

godot-rust v0.4 released!

Export groups, match_class, easy callables, run_deferred, generic PackedArray and much more!

godot-rust.github.io/dev/septembe...

Good morning Bay Area!

i was privileged to become american thanks to the h-1b visa. the program has its flaws, but adding a $100k "bribe" (which can be arbitrarily waived if you please the right people) is not the answer

True, but Rust has the key advantage of a strict compiler. Whatever the LLM produces you have higher confidence with Rust than with JS.

What a lovely little village! Would be a shame if it was just 0.02% the size of the world, right? 😅

#indiegames #indiedev #gamedev

This looks super slick

Allyon: Battlecry is a monster trainer where your creature actually understands you. Watch Flutterfox respond to natural voice commands—digging pits, displacing rocks with Bubble Blast, teleporting… We Need To Train! | Allyon #monstertamer #gaming #funny

Showing off our one & only Flutterfox with some training & combat! 🏋️ 🔥

youtube.com/shorts/13oIn...

Want to know more?

📱 Follow our socials for dev logs & clips:

- bsky.app/profile/clem...
- x.com/AllyonBattle...
- www.tiktok.com/@allyonbattl...
- www.instagram.com/allyonbattle...
- clementine.games

7/7

For relationships to blossom, play is as important as training or fighting. You can hang out with your Allyon at camp, talk to them, play games - it's a sandbox for both you and your Allyon.

6/7

For improvisation, battles must demand quick adaptation. So we made the environment highly dynamic, with Allyon powers that change it through side effects. Drench flames to create blinding steam clouds. Blast the ground beneath enemies to trap them in craters.

4/7

Preparation comes from training. You command during battles, but also train beforehand. Example: train your Allyon "when I call for help, teleport to me and build an ice wall." Then during battle, press one button and it happens - because you trained them to do it!

3/7

As I was saying…

Allyon is basically the game I always wanted to see from Pokemon. The appeal of monster trainer fantasy: 1) battles decided by preparation, improvisation and mutual trust, and 2) smart, magic pets you can nurture. These are core to our game.

2/7

Before I continue... links!

🎥 Video version of this thread on YT: youtu.be/ThOCM9TK_yo

❤️ Wishlist on Steam: store.steampowered.com/app/3987700/...

💬 Join our Discord to chat and snag playtests: discord.gg/AvyYjTER

We are building Allyon: Battlecry, the first monster trainer where creatures actually learn from you. It's closer to the Pokemon anime than games: Allyon understand your voice, learn bespoke behaviors, have personalities and memories, and form lasting bonds with you.

1/7

This is almost definitely true, but keep an eye out for the frequency, quality, and specificity of the data to gradually erode over the next 4 or more years. Our bureaucracy is huge and slow, which helps inoculate it against populist assault. Double edged sword, but right now, it's a plus.

Why do they love to use confusing nomenclature

Screenshot of a chat with Claude. User message: 'can you build an mcp for my ai-native b2b genai yc saas startup? pls'. Claude replies: 'No. What the fuck?' Below, a notice says 'Claude has ended this chat.' At the bottom, it shows 'Chat ended by Claude' with two buttons: 'Start new chat' and 'Give feedback'.

Flutterfox's Hydrostreak in action.

If you want to see more of our new upcoming game, Allyon, check us out below at our new website or our youtube handle where we post frequent new changes and improvements!

www.youtube.com/shorts/u69Nd...

#gamedev

Is this dependent on specific version of godot?

Ep 17: How Supercell is supporting AI-native innovation

Otto Söderlund, AI lead at Supercell, discusses his journey. We talk Supercell's AI initiatives. Otto also describes the AI Innovation Lab, which aims to explore radical ideas and support both external and internal projects.

Join us on the journey to build the world’s first true monster tamer — Allyon. In Allyon, your voice is the controller. Command monsters in real time. Watch them grow, adapt, and learn from you —… Best Duel of 2025? (Officially Unofficial Award) | Allyon #gamedev #monstertamer

www.youtube.com/shorts/mUg_c...

A peak into the combat system in our new game, Allyon.
#gamedev #monstertamer

It looks like somebody taught Flutterfox a new combat style!

If you want to see more of our new upcoming game, Allyon, check us out below at our new website or our youtube handle where we post frequent new changes and improvements!

clementine.games
www.youtube.com/@clementine-...

#gamedev

📢 Introducing Kaggle Game Arena: a new, open benchmark platform where top AI models compete in complex, strategic games in streamed match-ups. We're charting new frontiers for trustworthy AI evaluation and it begins with chess — a classic proving ground for system intelligence.

Ep 16: Examples of AI-Native Games & Mechanics

We follow-up last week's episode on definitions with some examples we've seen in the market and what we're looking forward to, or might see, in the near future.

This model is pretty sassy, later in the thinking trace it said:

Self-check: Am I being too pedantic? Nah—if someone asks for impossible things, it’s better to gently correct than make fake art that could confuse them.

Nature is scary!