@with_gene2626: wtf? 397b nvfp4 is a 3 or 4 spark setup... showing better benchmarks then all of the current open weight models? anyone…
Summary
Ornith-1.5, a family of open-source LLMs, is introduced with variants up to 397B MoE, achieving state-of-the-art performance among comparable models and rivaling Claude Opus in benchmarks.
View Cached Full Text
Cached at: 08/20/26, 12:55 AM
wtf? 397b nvfp4 is a 3 or 4 spark setup… showing better benchmarks then all of the current open weight models? anyone got this up and running yet? this looks absolutely amazing based on benchmarks. @mr_r0b0t have you tested the large 397b one of these?
Ornith (@ornith_): Aloha! 🌺Introducing Ornith-1.5, a family of open-source LLMs spanning 9B Dense, 35B MoE, and 397B MoE, trained with self-improving strategies.
It achieves state-of-the-art performance among open-source models of comparable size and delivers performance comparable to Claude Opus
Similar Articles
Ornith-1.5 open models launch in 397B, 35B, and 9 B sizes (2 minute read)
Ornith-1.5 has launched a family of open AI models in 397B, 35B, and 9B sizes, featuring a self-improvement loop and achieving competitive benchmarks against top models like Claude Opus 4.8.
@anvie: Tested Ornith-1.0-9B, and its impressive for a model of that size. I don't believe this is just 9B!
Ornith-1.0 is a family of open-source LLMs specialized for agentic coding, spanning sizes from 9B to 397B and achieving state-of-the-art performance among open-source models of comparable size.
@svpino: For the first time, I feel open-weight models are impossible to ignore. We are at a point where these models are compet…
Santiago (@svpino) highlights MiniMax-M2.7, a 230B open-weight model that rivals top proprietary models like Opus 4.6 and GPT-5.4, achieving 440+ tokens/s inference on SambaNova at low cost.
@sudoingX: i was running Ornith new 35b moe on llama.cpp with a Q4 quant, 4 bit, small, fast. it hit ~78 tok/s. then i swapped eng…
A 35B MoE agentic coding model called Ornith runs near lossless at FP8 on a single DGX Spark, achieving 3M token context and ~36 tok/s, with speculative decoding expected to boost speed further.
@AdinaYakup: This is impressive! Ornith is new, but every release makes an impact This time: - 397B reaches 86.1 on Terminal-Bench 2…
Ornith-1.5 releases a family of open-source LLMs from 9B to 397B parameters, achieving state-of-the-art performance among comparable models and offering multiple deployment-friendly formats.