# Ornith-1.5-9B-NVFP4 by Ornith: Open-Weight Model Source: https://savrn.com/models/ornith-1-5-9b-nvfp4 Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve Ornith-1.5-9B-NVFP4 (6.7B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them. | Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits | | --- | --- | --- | --- | --- | --- | | 16-bit | 13.5 GB | 16.1 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 8-bit | 6.7 GB | 8.1 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 4-bit | 3.4 GB | 4.0 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the [SAVRN Index](https://savrn.com/ai-index/pricing/gpus), read Oct 22, 2026. [Ornith-1.5-9B-NVFP4 on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/ornith-1-5-9b-nvfp4/gpus) ## SAVRN's Notes on Ornith-1.5-9B-NVFP4 One file, 8.8 GB, published 2026-08-26 under MIT: this is a ready-to-serve NVFP4 repack, so the generic precision table — 16.1 GB at 16-bit, 4.0 GB at 4-bit, one MI300X at $1.85 an hour — is a sizing sketch, not the spec. The shipped checkpoint is the 8.8 GB figure, sized to load on a single consumer-class card, and the 262,144-token context is the headline feature. The name says 9B while the page reads 6.7B parameters; confirm the checkpoint before benchmarking. MIT removes licensing friction entirely, including commercial use. There are no evaluations on file and no stated lineage beyond the Ornith publisher name, so this is a trust-but-verify download: attractive for a cheap long-context box, but run your own evals before trusting it with real traffic. SAVRN Research ## Model Card By Ornith, published under mit, revision 155f200d85ad. ### Ornith-1.5-9B Chirp Chirp! We are introducing Ornith-1.5, a major step toward building foundation models through end-to-end self-improvement. Ornith-1.5 extends Ornith-1.0 (which was developed on top of Qwen3.5 and Gemma4 with additional continued pretraining, mid-training, and post-training) by expanding the self-improvement loop from scaffold and rollout optimization to jointly optimizing task generation, scaffold construction, and solution rollouts. Rather than relying on a fixed set of human-curated tasks and manually designed harnesses, Ornith-1.5 continuously generates new training tasks, discovers effective strategies for solving them, and improves the policy through reinforcement learning. For more details on the task, harness, and rollout reward design, please refer to our [blog](https://ornith.ai/ornith_1_5.html). ### Ornith 1.5 9B This model card documents Ornith-1.5-9B, the most lightweight member of the Ornith-1.5 family — a 9B dense model designed for efficient single-GPU deployment, and edge-deployable on mobile devices via its quantized Ornith-1.5-9B-Mobile variant. #### Benchmarks [Read the full model card (1,429 words)](https://savrn.com/models/ornith-1-5-9b-nvfp4/card) ## Configuration Architecture Qwen3_5ForConditionalGeneration Context length (tokens) 262,144 Layers 32 Hidden size 4,096 Feed-forward size 12,288 Attention heads 16 Key/value heads 4 Head dimension 256 Vocabulary size 248,320 Model type qwen3_5 Quantization modelopt ## Identity and Version Repository ornith-ai/Ornith-1.5-9B-NVFP4 Publisher Ornith Task Text generation Modality Text Library transformers Parameters 6.7B parameters Languages Not stated by the source Revision 155f200d85ad58464571c77d5e1122ea5d419d7b First published 2026-08-26 Last updated 2026-08-26 ## Files and Weights 17 files, 8.8 GB in total. The weights are 1 file totalling 8.8 GB in safetensors. Weights1 file · 8.8 GB Configuration6 files · 68.3 KB Tokenizer4 files · 22.9 MB Documentation1 file · 26.8 KB Other3 files · 1.7 MB Repository2 files · 118.5 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model.safetensors | Weights | 8.8 GB | 1cdf915951ce | | config.json | Configuration | 38.5 KB | — | | generation_config.json | Configuration | 206 B | — | | hf_quant_config.json | Configuration | 27.6 KB | — | | preprocessor_config.json | Configuration | 390 B | — | | processor_config.json | Configuration | 1.2 KB | — | | video_preprocessor_config.json | Configuration | 385 B | — | | README.md | Documentation | 26.8 KB | — | | assets/ornith_9b_eval.png | Other | 724.3 KB | 5f13eb1febaa | | assets/ornith_logo.png | Other | 962.4 KB | 458ee0d85bae | | chat_template.jinja | Other | 7.6 KB | — | | .gitattributes | Repository | 1.7 KB | — | | .quant_summary.txt | Repository | 116.8 KB | — | | merges.txt | Tokenizer | 3.4 MB | — | | tokenizer.json | Tokenizer | 12.8 MB | 5f9e4d4901a9 | | tokenizer_config.json | Tokenizer | 16.7 KB | — | | vocab.json | Tokenizer | 6.7 MB | — | ## License and Download License mit Access Open weights, no gate Download size 8.8 GB [Download from Ornith](https://huggingface.co/ornith-ai/Ornith-1.5-9B-NVFP4) Released by Ornith through its official repository on Hugging Face. [Read the license](https://opensource.org/license/mit). ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 8.8 GB | | 16-bit | 13.5 GB | | 8-bit | 6.7 GB | | 4-bit | 3.4 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Questions About Ornith-1.5-9B-NVFP4 ### How much GPU memory does Ornith-1.5-9B-NVFP4 need? About 16.1 GB at 16-bit and 4 GB at 4-bit: the weights (6.7B parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run Ornith-1.5-9B-NVFP4 on? At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### Can I use Ornith-1.5-9B-NVFP4 commercially? Yes. Ornith-1.5-9B-NVFP4 is released under MIT License. The MIT License is a short permissive license. It permits commercial use, modification and redistribution, provided the copyright notice and permission notice are included. ### What is Ornith-1.5-9B-NVFP4's context length? 262,144 tokens, from the maximum position embeddings in its published configuration. ## Similar Models Model · Text generation ### [svd-safety-l2_remove40_swapgapiter_rankunit_b010](https://savrn.com/models/svd-safety-l2-remove40-swapgapiter-rankunit-b010) [Park](https://savrn.com/model-publishers/jeesup) A Llama-2-7b-chat checkpoint compressed with SVD-LLM to 60.0% of dense parameters, then edited by 10 of 10 rounds of iterative parameter-neutral swap selected by the gapiter rule (up to 0.1% of dense parameters per round; the full run's budget is 1.0%). This is a research artifact from a study of how SVD compression damages safety behaviour and which component-selection rule best repairs it. It is one cell of a grid over selection rules and budgets; it is not a general-purpose chat model. This checkpoint exists to measure safety/utility trade-offs under compression. Several arms in the grid are deliberately safety-degraded relative to Llama-2-7b-chat: compression alone raises attack-success… Open weights llama2 6.7B parameters 4,096 tokens transformers [View model](https://savrn.com/models/svd-safety-l2-remove40-swapgapiter-rankunit-b010) Model · Text generation ### [svd-safety-l2_jbbpure_ka1_a1p0_free_remove40](https://savrn.com/models/svd-safety-l2-jbbpure-ka1-a1p0-free-remove40) [Park](https://savrn.com/model-publishers/jeesup) A Llama-2-7b-chat checkpoint compressed with SVD-LLM to 60.0% of dense parameters, then given a 0.0% parameter budget of restored SVD components selected by the unknown rule. This is a research artifact from a study of how SVD compression damages safety behaviour and which component-selection rule best repairs it. It is one cell of a grid over selection rules and budgets; it is not a general-purpose chat model. This checkpoint exists to measure safety/utility trade-offs under compression. Several arms in the grid are deliberately safety-degraded relative to Llama-2-7b-chat: compression alone raises attack-success rate, and the point of the study is to quantify that and test recovery. Treat… Open weights llama2 6.7B parameters 4,096 tokens transformers [View model](https://savrn.com/models/svd-safety-l2-jbbpure-ka1-a1p0-free-remove40) Model · Text generation ### [Llama-2-7b-hf](https://savrn.com/models/llama-2-7b-hf) [Meta Llama](https://savrn.com/model-publishers/meta-llama) Llama 2 is a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters. This is the repository for the 7B pretrained model, converted for the Hugging Face Transformers format. Links to other models can be found in the index at the bottom. Note: Use of this model is governed by the Meta license. In order to download the model weights and tokenizer, please visit the website and accept our License before requesting access here. Meta developed and publicly released the Llama 2 family of large language models (LLMs), a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion… Access requested at publisher llama2 6.7B parameters transformers [View model](https://savrn.com/models/llama-2-7b-hf) Model · Text generation ### [finma-7b-nlp](https://savrn.com/models/finma-7b-nlp) [The Fin AI](https://savrn.com/model-publishers/thefinai) FinMA-7B-NLP is a financial large language model (LLM) developed as part of the PIXIU project. It is designed to understand complex financial language and concepts, and is fine-tuned to follow natural language instructions, enhancing its performance in downstream financial tasks. Specifically, FinMA-7B-NLP is trained only on the NLP tasks of the PIXIU dataset, making it specialized for tasks such as sentiment analysis, news headline classification, named entity recognition, and question answering. In addition to FinMA-7B-NLP, the PIXIU project includes two other models: FinMA-7B-full and FinMA-30B. You can use the FinMA-7B-NLP model in your Python project with the Hugging Face Transformers… Access requested at publisher mit 6.7B parameters transformers [View model](https://savrn.com/models/finma-7b-nlp) Model · Text generation ### [finma-7b-full](https://savrn.com/models/finma-7b-full) [The Fin AI](https://savrn.com/model-publishers/thefinai) FinMA-7B-full is a comprehensive financial large language model (LLM) developed as part of the PIXIU project. It is designed to understand complex financial language and concepts, and is fine-tuned to follow natural language instructions, enhancing its performance in downstream financial tasks. Specifically, FinMA-7B-full is trained with the full instruction data from the PIXIU dataset, covering both NLP and prediction tasks. This makes it a more comprehensive model capable of handling a wider range of financial tasks. In addition to FinMA-7B-full, the PIXIU project includes two other models: FinMA-7B-NLP and FinMA-30B. You can use the FinMA-7B-full model in your Python project with the… Access requested at publisher mit 6.7B parameters transformers [View model](https://savrn.com/models/finma-7b-full) Model · Text generation ### [deepseek-coder-7b-instruct-v1.5](https://savrn.com/models/deepseek-coder-7b-instruct-v1-5) [DeepSeek](https://savrn.com/model-publishers/deepseek-ai) Deepseek-Coder-7B-Instruct-v1.5 is continue pre-trained from Deepseek-LLM 7B on 2T tokens by employing a window size of 4K and next token prediction objective, and then fine-tuned on 2B tokens of instruction data. Here give some examples of how to use our model. This code repository is licensed under the MIT License. The use of DeepSeek Coder models is subject to the Model License. DeepSeek Coder supports commercial use. See the LICENSE-MODEL for more details. If you have any questions, please raise an issue or contact us at service@deepseek.com. Open weights other 6.9B parameters 4,096 tokens transformers [View model](https://savrn.com/models/deepseek-coder-7b-instruct-v1-5) ## Ornith [All models and datasets](https://savrn.com/model-publishers/ornith-ai) ## Versions - [155f200d85ad](https://savrn.com/models/ornith-1-5-9b-nvfp4/versions/155f200d85ad) · current 2026-09-18 ## Explore More - [All text generation models](https://savrn.com/models/tasks/text-generation) - [All models under mit](https://savrn.com/models/licenses/mit) - [Model comparisons](https://savrn.com/models/comparisons) - [The model directory](https://savrn.com/models) - [Open model prices by host](https://savrn.com/ai-index/pricing/open-models) ## Source - Repository metadata, read 2026-10-19. - [Hugging Face record](https://huggingface.co/ornith-ai/Ornith-1.5-9B-NVFP4) - [How the hub is built](https://savrn.com/model-hub/methodology)