Meet our third @MicrosoftAI model: MAI-Image-1 #9 on LMArena, striking an impressive balance of generation speed and quality Excited to keep refining + climbing the leaderboard from here! We're just getting started. https://
microsoft.ai/news/introduci
ng-mai-image-1-debuting-in-the-top-10-on-lmarena/
…
LLMS
-
Microsoft Debuts MAI-Image-1 Model Ranking Top 10 LMArena
By
–
-
Qwen 235B and Qwen3 Coder Now Available on Cerebras
By
–
Qwen 235B instruct, thinking, Qwen3 Coder all on Cerebras
-
Poe API Enables Tool Calling for Agentic Workflows
By
–
The Poe API now supports tool calling for all OpenAI, Anthropic, and Google models. This means you can use the Poe API to power agentic workflows directly with code or in applications like @KiloCode
, @roocode
, @cline
, @n8n_io
, Codex CLI and more. -
Cerebras AI Pricing Information Now Available Online
By
–
pricing is in the table: https://
cerebras.ai/pricing -
Cerebras Launches Self-Serve Inference at $10 Starting Price
By
–
🟧 Cerebras Inference self-serve is finally here 🟧
— Cerebras (@cerebras) 13 octobre 2025
– Pay by credit card starting at $10
– Run Qwen3 Coder, GPT OSS & more at 2,000+ TPS
– 20x the speed of GPU-based model providers
Go ahead. Melt our wafers.https://t.co/YfjxBGMl8i pic.twitter.com/FITWN8oVwCCerebras Inference self-serve is finally here – Pay by credit card starting at $10
– Run Qwen3 Coder, GPT OSS & more at 2,000+ TPS – 20x the speed of GPU-based model providers
Go ahead. Melt our wafers. http://
cloud.cerebras.ai -
Master AI in 2025: From Code to Career Development
By
–
Master AI in 2025
From code to career Build strong foundations Master ML & Deep Learning Learn LLMs & MLOps Deploy real AI projects Transition to AI careers AI is no longer an option — it's a key skill.
What's your next step? #AI #MachineLearning -
Model Quantization for Inference: Technical Questions
By
–
Thank you! Notably I didn't yet include model quantization for inference. I have questions 🙂
-
Hugging Face Datasets Critical Infrastructure for AI Model Training
By
–
Ty! huggingface work/infra/datasets are critical to projects like nanochat – to be accurate the source code of nanochat (e.g. at the $100 tier) is ~8KB of Python and ~30GB of fineweb/smoltalk.
-
QK Norm and Embedding Weight Sharing in Model Optimization
By
–
Very early on in the project I did a small run with/without QK norm and found that it helped. Same for the embedding weight sharing. I'll retry! I'm not tied to any details of the model and they weren't chosen any more carefully than a single run, I spent most of the time just
-
Advanced Llama Architecture: Rotary Embeddings and ReLU² MLP
By
–
> llama-like architecture
> dense transformer > rotary only (no positional embeddings) > qk norm > untied embedding/unembedding > norm after token embedding > relu² mlp > no biases in linears > no learnable rmsnorm params > mqa > logit softcap > optimizer =