- Size
- 780B parameterscounted from the safetensors header on the Hub
- Licence
- unknown, Apache License 2.0Our reading, check the licence: You can use, change and redistribute the model, including commercially. Keep the licence and any notices, and say which files you changed; you do not have to publish your changes. It includes a patent licence that ends for anyone who sues over patents in the work.
- On the Hub since
- 27 Aug 2026the date the repo was created
- Context length
- 1M tokensfrom the model's config
- Kind
- Base model (no base model declared on the Hub)HYV4ForCausalLM
- On Hugging Face
- tencent/Hy4-preview11k downloads in the last 30 days, 565 likes
Scores
- effort unknown35.0%surgehq.ai
Epoch's name: hy4-preview. Best closed model on this set: gemini-4-carbon_high, 56.4% (Google DeepMind). All 19 results
- effort unknown75.0%vals.ai
Epoch's name: hy4-preview. Best closed model on this set: claude-fable-5-1_max, 100.0% (Anthropic). All 85 results
Claimed by the maker Claimed
The model card has no claimed scores in a machine-readable form (the card's model-index), so none are shown. The maker may give scores in the card text: read the card.
Other views, which we link to and do not copy: look this model up on Artificial Analysis and the Arena leaderboard.
Independent scores from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Each row links its own source. What the badges mean.
Run it yourself
Memory needed = weights + context cache + 0.50 GB for the runtime. Context cache at 8k tokens: 0.69 GB, worked out from the model's config (tencent/Hy4-preview), at 16-bit precision. Choose a machine above to see a verdict for each format. Method
GGUF for llama.cpp, Ollama, LM Studio and others
6block/Hy4-preview-GGUF (converted by 6block), 543 downloads. Sizes are the real files on the Hub, all parts added up.
| Format | Size | Memory needed | Get it |
|---|---|---|---|
| IQ2_XS | 219.0 GBfile size, 39 parts | 220.1 GB219.0 + 0.69 + 0.50 | Commandsllama.cpp llama-server -hf 6block/Hy4-preview-GGUF:IQ2_XSDownload hf download 6block/Hy4-preview-GGUF --include "Hy4-preview-IQ2_XS-*-of-00039.gguf" --local-dir Hy4-preview-GGUF |
| Q2_K | 269.7 GBfile size, 39 parts | 270.8 GB269.7 + 0.69 + 0.50 | Commandsllama.cpp llama-server -hf 6block/Hy4-preview-GGUF:Q2_KDownload hf download 6block/Hy4-preview-GGUF --include "Hy4-preview-Q2_K-*-of-00039.gguf" --local-dir Hy4-preview-GGUF |
| IQ4_XS | 388.3 GBfile size, 39 parts | 389.5 GB388.3 + 0.69 + 0.50 | Commandsllama.cpp llama-server -hf 6block/Hy4-preview-GGUF:IQ4_XSDownload hf download 6block/Hy4-preview-GGUF --include "Hy4-preview-IQ4_XS-*-of-00039.gguf" --local-dir Hy4-preview-GGUF |
MLX for Apple silicon
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| MLX 4bitmlx-community/Hy4-preview-4bit | 403.6 GBfile size | 404.8 GB403.6 + 0.69 + 0.50 | Download page |
Original weights size from the safetensors header, for vLLM, SGLang, Transformers
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| BF16original weights | 1452.8 GBfile size | 1454.0 GB1452.8 + 0.69 + 0.50 |
hf download tencent/Hy4-preview --local-dir Hy4-previewOther conversions (3; sizes not read)
- tencent/Hy4-preview-FP8 FP8, by the maker, 15k downloads
- Oxmiq/Hy4-preview-NVFP4-W4A16 NVFP4, 471 downloads
- anemll/Hy4-preview-FlashMoE-STQ1_0 GGUF, 259 downloads
The Hub tags 5 quantised repos as conversions of this model; we read the most downloaded and show the maker's own first.
Run it by API
5 providers serve this model through the Hugging Face router. Prices are US dollars per million tokens. Precision is what the provider reports; a lower precision than the original weights can change answers. Live first, then cheapest output.
| Provider | In | Out | Context | Precision | Tokens/s | First token | Uptime 24h | Tools | JSON |
|---|---|---|---|---|---|---|---|---|---|
| DeepInfra | $0.83 | $2.50 | 1M | fp4 | 45 | 2.6 s | 100.0% | yes | yes |
| Novita | $0.83 | $2.50 | 977k | fp8 | 39 | 5.1 s | 100.0% | yes | yes |
| SiliconFlow | $0.83 | $2.50 | 1M | fp8 | 54 | 1.9 s | 100.0% | yes | yes |
| Tencentthe maker | $0.83 | $2.50 | 1M | fp8 | 30 | 3.2 s | 100.0% | yes | yes |
| Venice | $1.04 | $3.13 | 1M | fp4 | 45 | 1.4 s | 100.0% | yes | yes |
Through the router, with a Hugging Face token (the router bills your Hugging Face account). Change the name after the colon to pick another provider.
curl https://router.huggingface.co/v1/chat/completions \
-H "Authorization: Bearer $HF_TOKEN" -H "Content-Type: application/json" \
-d '{"model": "tencent/Hy4-preview:deepinfra", "messages": [{"role": "user", "content": "Hello"}]}'Not built yet: one key for every open model, with BlockRun, and our own tested-quality score per provider. Until then, precision and measured speed are the best guide to a provider here.
Planned for this page Not built yet
- Try it in private chat
- A link to Solana AI's confidential chat for the models it serves. This model is not listed as served there.
- Fine-tune it
- Fine-tuning a few base models from this page and publishing the result here.
- Creators and earnings
- Pages for people who publish fine-tunes and adapters of this model, and what they earn.
- Fund and bounties
- Bounties for new open models, paid from a public fund. How the fund will work.
- Contribute compute
- How to give compute to open model work, with Akash.
Model data from the Hugging Face Hub 19 Nov 2026, 13:43 UTC. The repo. Found a mistake on this page? Tell us.