Model Releases
My feed has been hijacked this week by frontier model influencers who say they've had 5.6 Sol and Fable for 'months'. I think this tells an …
My feed has been hijacked this week by frontier model influencers who say they've had 5.6 Sol and Fable for 'months'. I think this tells an inaccurate story the field. From the outside it makes AI pro
My feed has been hijacked this week by frontier model influencers who say they've had 5.6 Sol and Fable for "months". I think this tells an inaccurate story the field. From the outside it makes AI progress feel like a spectator sport where a small group gets the model early and everyone else waits, then we all accept whatever price these labs decide to charge. I get excited about releases like this one between us and NVIDIA because they show how the pace of AI development is NOT just decided by the release cadence of SOTA models. Together we tuned our open Deep Agents harness for Nemotron 3 Ultra and scored 0.86 on our eval suite for 4.48... the next closest model by performance cost almost 10x more, 43.48. So amid all the closed model hype, remember that open models are rapidly improving and so are the systems that let us to maximize their performance. To me this is what democratizing AI looks like, because devs gain control over the systems they are building, agent progress doesn't wait on a handful of labs, and teams are given better options than to get financially rinsed for perf gains that aren't necessary for their agents to succeed. Introducing the NemoClaw Deep Agents Blueprint, a reference architecture for building open agent systems developed with @NVIDIA ✅ A fully open stack enterprises can own and customize ✅ Benchmark-leading performance ✅ Over 10x lower inference costs Blog: http://www.langchain.com/b…
Source: Harrison Chase (X) | 2026-07-09