Before LangSmith LLM Gateway: An agent processes a request that includes a SSN. It now sits in LLM provider logs, in trace data, + possibly in downstream systems that consumed the response. With LangSmith LLM Gateway: Data is redacted from requests before it hits a model or
LLMS
-
AI Model Comparison: Opus 4.8 vs. GPT-5.5 Trajectory
By
–
Opus 4.8 is clearly a strong model, but my impression is that Anthropic is increasingly playing catch-up with OpenAI rather than setting the pace. It feels like GPT-5.5 has shifted the benchmark again, and if OpenAI keeps this trajectory, GPT-5.6 could very plausibly become the
-
LiquidAI Reasoning Model: 1B Parameters, On-Device Performance
By
–
LiquidAI has built a tiny reasoning model that achieves useful performance for on device use cases with only 1B active parameters
-
Claude Opus 4.8 Hermes Agent Performance Update
By
–
Claude Opus 4.8 in Hermes Agent is on a completely different level. My Agent team just got much more smarter in just one update.
-
General Compute Builds Efficient AI Inference Cloud Infrastructure
By
–
@TimFernholz at @TechCrunch breaks down how General Compute is building its inference cloud with SambaNova, and why faster, more efficient inference infrastructure is becoming critical for the next wave of AI. Featuring insights from General Compute CEO @FPuklowski and CTO
-
Deep Agents v0.6 Introduces ContextHubBackend for Agent File Management
By
–
New in Deep Agents v0.6: ContextHubBackend A versioned home for the files that power agent behavior, backed by LangSmith Context Hub, enabling context improvements from one run to the next.
-
Anthropic to release mythos-class models in coming weeks
By
–
new from me: anthropic is widely releasing "mythos-class models" in the coming weeks. the company had forecast this last week, but today tucked the timing into the very bottom of a blog post about claude opus 4.8.
-
TaH: Skipping extra reasoning steps makes AI smarter
By
–
Your AI model is thinking too much, and skipping a few steps makes it smarter! Tsinghua University, Infinigence AI, and Shanghai Jiao Tong University introduce Think-at-Hard (TaH). Instead of looping every token through extra reasoning, TaH uses a lightweight decider to skip
-
AI Model Opus 4.8 Calibration and User Feedback Request
By
–
We put a lot of work into calibrating thinking effort for Opus 4.8. As you're trying out the model, if you do run into any examples of it still over/under thinking, please flag it to us!
-
ByteDance Open-Sources Capable Multimodal AI Model BAGEL
By
–
ByteDance just open-sourced one of the most capable multimodal models out there.
— Chubby♨️ (@kimmonismus) 28 mai 2026
BAGEL does image generation, editing, style transfer, and visual understanding – all in a single 7B parameter model. Apache 2.0 licensed!
One model. No switching between specialized tools.
Amazing https://t.co/r4jj4QHTzMByteDance just open-sourced one of the most capable multimodal models out there. BAGEL does image generation, editing, style transfer, and visual understanding – all in a single 7B parameter model. Apache 2.0 licensed! One model. No switching between specialized tools.
Amazing