Text-to-Image Arena helps you find image generation models with granular categories. Now, we’re bringing that same level of detail to image editing by launching categories in Single and Multi-Image Edit Arenas.

With millions of real user comparisons, Arena…

We analyzed >3M votes to see if our 7 categories were truly independent. The answer? Not exactly.
- Photorealistic + Portraits overlap 64%.
- Commercial Design + Text Rendering overlap 36%.
- Photorealistic and Art overlap under 1% — they're different jobs.

For more details, check out the full leaderboard: https://arena.ai/leaderboard/image-edit

Today we’re launching AutoEval: a new evaluation methodology that ranks models using reward models based on millions of real Arena user preferences.

Highlights:
- High-quality evaluation signals calibrated on real preference data
- Strong alignment with…

How does AutoEval produce Arena Score?

For a new model to be evaluated, we generate responses on prompts sampled from live Arena evaluations. A pointwise reward model on millions of pairwise human votes to score each response and converts score differences…

We train our reward model to handle challenges like preference shifts, increasingly subtle model differences, ambiguous tie votes, and longer and more context-dependent interactions. To evaluate it fairly, we use a temporal holdout: train on historical votes…

AutoEval scores provide an early signal for comparing candidate model checkpoints before full live results become available.

We compared AutoEval’s early ordering with the final live leaderboard. When two models were separated by at least 10 Arena points…

The same approach extends beyond text. Across vision, image generation, and code arena, we train modality-specific reward models on live Arena preferences.

For example, on Text-to-Image, our reward model was trained on more than 3 million preference pairs.…

Read more about AutoEvals on our blog: http://arena.ai/blog/autoeval-scores

Inkling-Small (@thinkymachines) debuts at rank ~#88 (1431 pts, AutoEval) in Text Arena and ~#21 among open.

This makes it one of only five US models in the top 25 open models overall.

Note: this is an early AutoEval score, in which a Reward Model trained…

Dive into the Inkling-Small AutoEval scores on Text Arena at: http://arena.ai/leaderboard/text

Read more about AutoEval scores and our methodology on our blog: http://arena.ai/blog/autoeval-scores

⛏️ Block 960,289 just signaled BIP-110!
15 of 674 blocks this period — 2.23%.
#BIP110 #Bitcoin [period 476]
https://bip110monitor.com?data=eyJwY3QiOjIuMjMsInNpZyI6MTUsInRvdGFsIjo2NzQsInBlcmlvZCI6NDc2LCJzeW5jZWQiOnRydWUsImJpcE51bSI6IjExMCJ9.zqjSWUbcLXoHTC…

All of a sudden, every major tech company wants your face on your device, and that is suspicious. We also look at more surveillance expansion, and other articles in privacy, security, and technology.

Gemini Robotics 2 includes three models, but only one is publicly available right now.

The U.S. wants Bitcoin as a strategic asset but resists its censorship resistance, privacy tools, and growing role as real financial infrastructure.

🟢Since December, the supply held for more than 1 year keeps increasing.

💥 This has recently accelerated, with the monthly average now at +217,000 Bitcoin.

This is a positive development, indicating that many holders who bought a year ago, and therefore…

BUKELE: "We never escaped the 2008 financial crisis. We just shifted the pain."

TREASURY SECRETARY SCOTT BESSENT QUOTES SATOSHI

"America will lead or America won’t. It’s not more complicated than that. I believe Satoshi once said it best: “If you don’t believe me or don’t get it, I don’t have time to try to convince you, sorry.”

Day 1 of BTC++ Consensus Edition 👀

bitcoin++ payments edition coming to berlin this fall -- october 1-3

agents welcome. have them buy u your tickets now

https://btcpp.dev/berlin26