How @cognition /
@windsurf new SWE-1.5 model was built (probably), based on piecing together bits of information that was shared. Base model by: @Zai_org possibly GLM-4.6
RL Training: @nvidia on 'thousands' GB200 NVL72
Inference: @cerebras at 950 tks/sec
LLMS
-
SWE-1.5 Model Architecture: Base Model, RL Training, Inference
By
–
-
Cartesia Sonic-3 Launches on Poe with 100 Voices
By
–
Congrats to the @Cartesia_ai team and thanks for bringing Sonic-3 to Poe.
— Poe (@poe_platform) 31 octobre 2025
You can use it at https://t.co/OIC6uRjktZ, on our apps, or with the Poe API, with full support for the model’s 100 new voices, 42 languages, and human-like expressiveness. https://t.co/B4xjOdbjkiCongrats to the @cartesia team and thanks for bringing Sonic-3 to Poe. You can use it at https://
poe.com/Cartesia-Sonic
-3.0
…, on our apps, or with the Poe API, with full support for the model’s 100 new voices, 42 languages, and human-like expressiveness. -
Language Models Struggle With High School Math Fundamentals
By
–
Language models perform poorly on high-school math? You don't want to hear this, but the problems started in grade-school. The moment we (collectively) found acceptable that mid-tier models could score only 75%-85% on a GSM test set of 1.32k straightforward problems…
-
Part 4: AI Evaluations with Clefourrier
By
–
Wow, awesome! But why stop at a trilogy when you could do a part 4 about evaluations with @clefourrier
? -
Google Reports Strong Quarterly Growth Driven by Gemini Models and TPU
By
–
Great to see major increases in many metrics (love all the usage of the Gemini app!), many of them driven by Gemini models and our TPU hardware. Congrats to all Googlers on a great quarter!
-
OpenAI Launches Aardvark Security Research Agent
By
–
OpenAI is launching a new Security Research Agent, "Aardvark", that can identify and resolve vulnerabilities in code. Aardvark is currently available in private beta and powered by GPT-5 and Codex.
-
Stanford Method Detects Copied Fine-Tuned AI Models
By
–
Someone stole your model & u can’t prove it? This Stanford paper just showed that you can find out if a model is a copy or finetuned based on your model with just its generated text So if someone yoinks DeepSeek-v3.2 and finetunes it, it’ll leave statistical traces!
-
Cerebras Introduces Code Platform for AI Development
By
–
https://
cerebras.ai/blog/introduci
ng-cerebras-code
… -
Faster Coding Model Pricing Efficiency Questions
By
–
The speed of a faster coding model is worth it, but it seems mis-priced. C1 gobbles through files, reasons more, expect extra feedback to reach similar place as slower model do with less of everything. Intuitively it feels more expensive "the fast way" with current pricing.
-
Google AI Studio Adds Logs and Datasets for AI Output Assessment
By
–
Google AI Studio got a new Logs and Datasets section to let users assess the quality of AI outputs.