Oriel — AI & technology worth noticing, every day.

ornith-ai/Ornith-1.5-122B

AI model · https://huggingface.co/ornith-ai/Ornith-1.5-122B

First seen by Oriel on .

AI summary

The Ornith-1.5-122B model card presents a 122B MOE reasoning model built through end-to-end self-improvement, with tool use, long-context support, and single-GPU deployment.

  • Ornith-1.5 jointly optimizes task generation, scaffold construction, and solution rollouts through reinforcement learning instead of relying on fixed human-curated tasks.
  • The model supports OpenAI-compatible serving through Transformers ≥ 5.8.1, vLLM ≥ 0.19.1, or SGLang ≥ 0.5.9.
  • Its 262,144-token context can be extended to roughly 1M tokens with YaRN, although static scaling may reduce quality on shorter inputs.
  • As a 122B MOE model, it serves on a single 80GB GPU and supports tool calling through standard OpenAI-compatible APIs and agent frameworks.
  • Reported results include 84.500 on SWE-bench Verified(non-hack), 81.200 on TerminalBench 2.1, and 78.500 on BrowseComp.

#large language models #reinforcement learning #moe #agentic ai

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Likes

Total likes reported by the source.

  • Hugging Face Models: 197, latest observation ; +88 over observed period since

Downloads (30d)

Rolling 30-day downloads. A change in this value is not new downloads during the observed period.

  • Hugging Face Models: 76,295, latest observation ; +75,553 over observed period since

Aliases

SourceExternal IDURL
Hugging Face Modelsornith-ai/Ornith-1.5-122Bhttps://huggingface.co/ornith-ai/Ornith-1.5-122B Read discussion