NVIDIA Nemotron™ is a family of open models with open weights, training data, and recipes, delivering leading efficiency and accuracy for building specialized AI agents.
Hemmingway-1.5
HemmingwayAI·Instruct·released 24 Oct 2026
ollama run hf.co/bartowski/HemmingwayAI_Hemmingway-1.5-GGUF:Q4_K_MThe undetectable AI. Writing indistinguishable from a human's. 27B parameters, open weights, free for non-commercial use.
How much VRAM does Hemmingway-1.5 need?
At Q4_K_M Hemmingway-1.5 needs about 18 GB with an 8K context (18 GB at Q4, 55 GB at full precision). Pick your GPU to see what fits and how fast it runs.
Needs: ≈20 GB · Q4_K_M · 8K context — Pick a GPU to check the fit.
Advanced: context 8K · KV cache 16-bit
Smallest GPU for each quant and context (KV cache 16-bit)
| Quantisation | 8K | 16K | 32K | 64K | 128K | 256K |
|---|---|---|---|---|---|---|
| Q3_K_M | ||||||
| Q4_K_M | ||||||
| Q5_K_M | ||||||
| Q6_K | ||||||
| Q8_0 | ||||||
| FP16 |
≤16 GB card24–32 GB48–96 GBmore
| Quantisation | Total |
|---|---|
| Q2_KHeavy lossSmallest, noticeably worse quality | ≈13 GB |
| Q3_K_MNoticeable lossSmall, some quality loss | ≈16 GB |
| Q4_K_MrecommendedSmall lossRecommended default for local use | ≈20 GB |
| Q5_K_MNear-losslessNear-lossless for most tasks | ≈23 GB |
| Q6_KNear-losslessPractically lossless | ≈26 GB |
| Q8_0LosslessLossless in practice, twice the size of Q4 | ≈31 GB |
| FP16 / BF16LosslessFull precision, as released | ≈56 GB |
Estimates for 8K tokens of context: weights + KV cache + runtime overhead. Speed assumes memory-bound decoding at your GPU’s bandwidth. Pick a GPU above and the fit column checks every row against it. Weights marked “file” use the real GGUF download size; “est.” rows are computed from the parameter count.
How to pick a quant
Take the biggest quant that fits with room left for context. Q4_K_M is the usual sweet spot. Below Q4 quality drops fast, above Q6 you mostly pay memory for nothing. Big models (70B+, large MoE) tolerate low quants better than small ones.
- LosslessF16, Q8_0
- Near-losslessQ6_K, Q5_K_M
- Small lossQ5_K_S, Q4_K_M, IQ4_NL, IQ4_XS, Q4_K_S
- Noticeable lossQ4_0, Q3_K_L, Q3_K_M, IQ3_M
- Heavy lossIQ3_XXS, Q3_K_S, IQ2_M, Q2_K
Typical quality loss, not measured per model. Tiers come from KL divergence against FP16 on Llama 3 8B: llama.cpp perplexity README: LLaMA 3 8B scoreboard (KLD and ΔPPL vs FP16, Wikitext-2); llama.cpp quantize tool: ΔPPL per quant type @ Llama-3-8B
How much context fits?
At 8K context, Q4_K_M needs ≈20 GB(16 GB weights + 2.0 GB KV cache + 1.3 GB overhead). Pick a GPU above to see how much context fits.
Weights + overheadKV cache
Can the Tesla P40 run Hemmingway-1.5? →
Speed vs quality per quant
Smaller quants run faster and fit in less memory, but lose quality.
Fits in VRAMTight fitCPU offload✕ won't runQuality is typical KL divergence vs FP16, not measured for this model.
Real-world speed of Hemmingway-1.5
Tokens per second measured by people running it at home. A GPU shows a median once 3 reports agree on the same setup. Reports far from our estimate wait for a review. Measure yours with llama-bench →
No medians yet. Be one of the first 3 to report a setup.
All reports
Run Hemmingway-1.5 locally
--quantQ4_K_Mdefault; pick your GPU to tailor itbartowski/HemmingwayAI_Hemmingway-1.5-GGUF
ollama run hf.co/bartowski/HemmingwayAI_Hemmingway-1.5-GGUF:Q4_K_M llama-server -hf bartowski/HemmingwayAI_Hemmingway-1.5-GGUF:Q4_K_MDownloads the GGUF on first run. Installed through llama.app? The same flags work with `llama serve`.
lms get bartowski/HemmingwayAI_Hemmingway-1.5-GGUF@Q4_K_MThen load it from the LM Studio app or with `lms load`.
pip install mlx-lm
mlx_lm.generate --model NeosQ/Hemmingway-1.5-MTPLX-Q4 --prompt "Hello"Use `mlx_lm.chat` with the same --model for an interactive chat.
mlx_lm.server --model NeosQ/Hemmingway-1.5-MTPLX-Q4 --port 8080API on http://localhost:8080/v1.
vllm serve HemmingwayAI/Hemmingway-1.5 vllm serve cyankiwi/Hemmingway-1.5-AWQ-INT4services:
vllm:
image: vllm/vllm-openai:latest
command: HemmingwayAI/Hemmingway-1.5
ports:
- "8000:8000"
ipc: host
volumes:
- ~/.cache/huggingface:/root/.cache/huggingface
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]API on port 8000.
sglang serve HemmingwayAI/Hemmingway-1.5 --host 0.0.0.0 --port 30000services:
sglang:
image: lmsysorg/sglang:latest
command: sglang serve HemmingwayAI/Hemmingway-1.5 --host 0.0.0.0 --port 30000
ports:
- "30000:30000"
ipc: host
volumes:
- ~/.cache/huggingface:/root/.cache/huggingface
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]API on port 30000.
from transformers import pipeline
pipe = pipeline("text-generation", model="HemmingwayAI/Hemmingway-1.5", dtype="auto", device_map="auto")
messages = [{"role": "user", "content": "Explain self-hosting in one paragraph."}]
out = pipe(messages, max_new_tokens=512)
print(out[0]["generated_text"][-1]["content"])Quantised versions and downloads
| Repository | Format | Quant | Size |
|---|---|---|---|
| bartowski/HemmingwayAI_Hemmingway-1.5-GGUF | GGUF | IQ2_XXS | 8.9 GB |
| bartowski/HemmingwayAI_Hemmingway-1.5-GGUF | GGUF | IQ2_XS | 9.1 GB |
| bartowski/HemmingwayAI_Hemmingway-1.5-GGUF | GGUF | IQ2_S | 9.7 GB |
| bartowski/HemmingwayAI_Hemmingway-1.5-GGUF | GGUF | IQ2_M | 11 GB |
| bartowski/HemmingwayAI_Hemmingway-1.5-GGUF | GGUF | Q2_K | 11 GB |
| bartowski/HemmingwayAI_Hemmingway-1.5-GGUF | GGUF | IQ3_XXS | 12 GB |
| bartowski/HemmingwayAI_Hemmingway-1.5-GGUF | GGUF | Q3_K_S | 13 GB |
| bartowski/HemmingwayAI_Hemmingway-1.5-GGUF | GGUF | IQ3_XS | 13 GB |
| bartowski/HemmingwayAI_Hemmingway-1.5-GGUF | GGUF | Q3_K_M | 13 GB |
| bartowski/HemmingwayAI_Hemmingway-1.5-GGUF | GGUF | Q3_K_L | 14 GB |
About Hemmingway-1.5
The undetectable AI. Writing indistinguishable from a human's. 27B parameters, open weights, free for non-commercial use. hemmingway.io → · Model card → · Try it → · API → · API docs → · Mac, Windows and Android apps → · Code → · Product Hunt → Our other models: Hemmingway-1, our first model · Hemmingway-1-Unlocked, the uncensored adapter for Hemmingway-1, adults only Hemmingway-1.5 passes the world's best AI detectors 90% of the time. On GPTZero it passes 9 times in 10, at every thinking level. Real human writing passes the same detectors 86% of the time. Fable, GPT-6 and Gemini get caught most of the time. We had it write 1,848 texts across 13 kinds of writing, at all four thinking levels (off, low, medium and extra high), and ran each one through the detectors once. No picking, no retries. 90% came back human. We ran 1,670 texts that people wrote through the same detectors. 86% passed. Hemmingway-1.5 passes more often than they do. Hemmingway-1, the model it is built on, passed 7%. Thinking off, low, medium or extra high, Hemmingway-1.5 passes GPTZero 9 times in 10. We gave Hemmingway-1.5 and the biggest models the same 80 everyday asks, the asks from CommunicationBench, and ran every answer through the same detectors. One of the asks: a recruiter offered a senior PM role at a low salary, and the user wanted "a polite reply that says thanks but no thanks unless the budget changes". Hemmingway-1.5's reply read as human:
Full model card on Hugging Face →Alternatives to Hemmingway-1.5
All alternatives →Versioning notice: This card describes the current OTel 2.0 OSFT checkpoint. The model weights may be updated over time. For reproducible evaluation or production deployment, pin a specific model revision, checkpoint hash, or release tag rather than relying on the floating latest version.
Aloha! 🌺 Today, we are releasing Ornith-1.0, a self-improving family of open-source models for agentic coding.
Laguna XS 2.1 is a 33B total parameter Mixture-of-Experts model with 3B activated parameters per token designed for agentic coding and long-horizon work on a local machine.
👋 Join our Discord community. 📖 Check out the GLM-4.7 technical blog , technical report(GLM-4.5) . 📍 Use GLM-4.7-Flash API services on Z.ai API Platform. 👉 One click to GLM-4.7 .
Nemotron-3-Nano-30B-A3B-BF16 is a large language model (LLM) trained from scratch by NVIDIA, and designed as a unified model for both reasoning and non-reasoning tasks. It responds to user queries and tasks by first generating a reasoning trace and then concluding with a final response.
Hemmingway-1.5 FAQ
- How much VRAM does Hemmingway-1.5 need?
- About 18 GB at Q4_K_M with an 8K context. Q8_0 needs about 30 GB and full precision about 55 GB. Longer contexts need more memory for the KV cache.
- Can Hemmingway-1.5 run on the Tesla P40?
- Yes. It fits at Q4_K_M (about 20 GB), at an estimated 12 tokens/s. It fits comfortably on the Radeon RX 7900 XTX, Instinct MI50 32GB, RTX 5090.
- Can I use Hemmingway-1.5 commercially?
- No. It is released under CC BY-NC 4.0. Non-commercial use only, with attribution.
- What is the context length of Hemmingway-1.5?
- 262,144 tokens (256K). The full window needs far more memory than an 8K context.
- How do I run Hemmingway-1.5 with Ollama or llama.cpp?
- With Ollama: ollama run hf.co/bartowski/HemmingwayAI_Hemmingway-1.5-GGUF:Q4_K_M. With llama.cpp: llama-server -hf bartowski/HemmingwayAI_Hemmingway-1.5-GGUF:Q4_K_M, which starts an OpenAI-compatible server.
Rate this model
Data from Hugging Face, the model card and public leaderboards. Memory and speed figures are estimates; real numbers depend on the runtime, context length and drivers. Always check the licence yourself before commercial use.
0 comments