That Haiku number is the one worth sitting with. Claude Haiku scores 39% on SWE-Bench Pro. On DeepSWE, where it can't coast on contaminated data or exploit the test environment, it scores zero. Not low. Zero. That's not a model that dropped in performance. That's a model that
GENERATIVE AI
-
MP-MoE: Diverse Expert Routing Boosts Large Language Models
By
–
What if your AI model’s experts weren’t just the smartest, but the most diverse? Researchers from Renmin University, Huawei, and Tianjin University introduce MP-MoE: a routing method that selects diverse experts using co-occurrence patterns. Result: 1-3% boost in LLM
-
Get 1 Billion Free LLM Tokens Monthly by Combining Free Tiers
By
–
Want ONE BILLION free LLM tokens a month without juggling a dozen different APIs? Now you can *legally* unlock that massive inference capacity by combining the free tiers of Google, Groq, SambaNova, Mistral, and GitHub Models. The only problem is the headache of managing all
-
Google Genie experiment generates virtual worlds from Maps locations
By
–
ICYMI : Users with access to Google Genie experiment now can use locations from Google Maps to generate virtual worlds. “Golden Gate Bridge”
-
Paper: Differences Between AI and Human Narrative Styles
By
–
There is a lot being written about the stylistic tells of AI writing (em-dashes, etc.) but this paper looks at AI narrative tells Fascinating differences between AI & human narrative, and asking AI to write in different styles doesn't do much to change it https://
arxiv.org/abs/2604.03136 -
Krea 2 AI Image Generator Now Available on Replicate
By
–
Krea 2 from @krea_ai is available on Replicate. Generate high-fidelity, creative images with aesthetics first in mind.
-
BIGAI’s NPR enables parallel reasoning in LLMs via self-distilled RL
By
–
What if LLMs could think in multiple directions at once, not just step by step? BIGAI introduces NPR: a teacher-free framework that lets LLMs self-evolve genuine parallel reasoning. Instead of emulating sequential logic, it uses self-distilled reinforcement learning and a
-
Looping transformer layers boosts AI language model efficiency
By
–
What if you could make AI language models smarter by reusing the same layers over and over? Researchers from KAIST, KRAFTON, and UC Berkeley present LoopMDM（Looped Diffusion Language Models）. They selectively loop early-middle transformer layers in masked diffusion models—no
-
Cursor’s frontier model with 100x fewer resources than Google
By
–
it is wild that Cursor trained a model closer to the frontier than Google with 100x fewer people and (guessing) ~100x less compute i am surprised this was even possible. also praying for the Gemini comeback ofc
-
Raising capital in 2024 for Llama 3, now dead on arrival
By
–
Imagine having raised in 2024 and spent the capital in doing this for llama 3 and putting that out on the market now Dead on arrival.