Ripple-65B Preview NVFP4 (Experimental)

Ricursive Intelligence introduces Ripple-65B Preview, a 65B parameter dense Transformer model pretrained on 12 trillion tokens with support for up to 1M context length. Ripple-65B Preview delivers strong results on agentic coding, scoring 80.52 on TerminalBench 2.1 and surpassing significantly-larger US open-weight models such as Reflection Beam 501B (80.1), Meta Muse Spark Open 816B (76.4), Thinking Machines Inkling 975B (63.8), and NVIDIA Nemotron 3 Ultra 550B (56.4).

Our Preview release includes an instruction-tuned model in BF16 and NVFP4 numerics, and a BF16 base model. This repo contains the NVFP4 instruction-tuned checkpoint.

This is a conservative quantization of the BF16 checkpoint, using NVFP4 for the MLP layers while keeping attention in BF16.

Citation

@misc{ricursive2026ripple65b,
      title={Ripple-65B: Pushing the Frontier of Local Intelligence},
      author={Ricursive Intelligence},
      year={2026},
}

Contact

For questions and comments, please contact us at info@ricursive.com. To learn more, please visit our website.

env CUDA_DEVICE_ORDER=PCI_BUS_ID CUDA_VISIBLE_DEVICES=0,1 CUDA_HOME=/usr/local/cuda-13.2 PATH=/usr/local/cuda-13.2/bin:/usr/local/bin:/usr/bin:/bin SGLANG_ENABLE_TP_MEMORY_INBALANCE_CHECK=false /home/puretensorai/venvs/ripple-sgl/bin/python /home/puretensorai/bench/ripple/serve_ripple_fp4.py /mnt/storage/models/ripple/Ripple-65B-Preview-NVFP4 /mnt/storage/models/ripple/Ripple-65B-Preview-DFlash2 --host 127.0.0.1 --port 8673 --context-length 32768 --tp-size 2 --speculative-num-draft-tokens 16

python -m sglang.launch_server (via serve_ripple_fp4.py: registers the vendor RicursiveForCausalLM plugin from ricursiveAI/Ripple-65B-Preview-NVFP4 inference/local + ripple parsers) --model-path ricursiveAI/Ripple-65B-Preview-NVFP4 --trust-remote-code --dtype bfloat16 --kv-cache-dtype bfloat16 --tp-size 1 --attention-backend flashinfer --fp4-gemm-backend flashinfer_cutlass --flashinfer-autotune-skip-ops fp8_gemm --context-length 32768 --mem-fraction-static 0.85 --max-running-requests 1 --cuda-graph-bs-decode 1 --cuda-graph-backend-prefill disabled --tool-call-parser ripple --reasoning-parser ripple --chat-template chat_template.jinja --speculative-algorithm DFLASH --speculative-draft-model-path ricursiveAI/Ripple-65B-Preview-DFlash2 --speculative-draft-model-quantization unquant --speculative-num-draft-tokens 16 --speculative-draft-window-size 16384 --speculative-draft-attention-backend triton (env SGLANG_FLASHINFER_USE_PAGED=1 FLASHINFER_USE_CUDA_NORM=1 CUDA_HOME=cuda-13.2)

ricursiveAI/Ripple-65B-Preview-NVFP4

Ripple-65B Preview NVFP4 (Experimental)

Citation

Contact

为此模型做种