Hy4-preview
Tencent Hy4 preview is Tencent Hy’s open-source large language model, featuring 770B total parameters, 49B active parameters, and a context window exceeding 1 million tokens.
text → text · made by Tencent Hunyuan
Sold by 22 ways
| Seller | Lane | Rate |
|---|---|---|
| Nous Research aggregator | standard | $0.42per Mtok in$1.25per Mtok out$0.021per Mtok cachedinference-api.nousresearch.com · read 2026-09-23 |
| OpenRouter 8 ways | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cached | |
| OpenRouter aggregatornot on the seller's list today | makora/fp8 | $0.8per Mtok in$2.48per Mtok out$0.038per Mtok cachedopenrouter.ai · read 2026-10-16 |
| OpenRouter aggregator | standard | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedopenrouter.ai · read 2026-09-20 |
| OpenRouter aggregator | atlas-cloud/fp8 | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedopenrouter.ai · read 2026-11-01 |
| OpenRouter aggregator | cohere | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedopenrouter.ai · read 2026-11-01 |
| OpenRouter aggregator | deepinfra/fp4 | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedopenrouter.ai · read 2026-10-24 |
| OpenRouter aggregator | gmicloud/bf16 | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedopenrouter.ai · read 2026-11-01 |
| OpenRouter aggregator | novita/fp8 | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedopenrouter.ai · read 2026-11-01 |
| OpenRouter aggregator | siliconflow/fp8 | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedopenrouter.ai · read 2026-11-01 |
| OpenRouter aggregator | tencent/fp8 | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedopenrouter.ai · read 2026-11-01 |
| Inference.net 0 ways | $0.83per Mtok in$2.5per Mtok out$0.04per Mtok cached | |
| Inference.net aggregatornot on the seller's list today | standard | $0.83per Mtok in$2.5per Mtok out$0.04per Mtok cachedapi.inference.net · read 2026-10-02 |
| AtlasCloud aggregator | standard | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedapi.atlascloud.ai · read 2026-10-28 |
| DeepInfra aggregator | standard | $0.83per Mtok in$2.5per Mtok outapi.deepinfra.com · read 2026-09-27 |
| Hugging Face Inference Providers 3 ways | $0.83per Mtok in$2.5per Mtok out | |
| Hugging Face Inference Providers aggregator | standard | $0.83per Mtok in$2.5per Mtok outmodels.dev · read 2026-11-01 |
| Hugging Face Inference Providers aggregator | deepinfra | $0.83per Mtok in$2.5per Mtok outrouter.huggingface.co · read 2026-11-01 |
| Hugging Face Inference Providers aggregator | novita | $0.83per Mtok in$2.5per Mtok outrouter.huggingface.co · read 2026-09-20 |
| Novita AI aggregator | standard | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedapi.novita.ai · read 2026-09-20 |
| Requesty aggregator | tencent | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedrouter.requesty.ai · read 2026-09-30 |
| Vercel AI Gateway aggregator | standard | $0.83per Mtok in$2.5per Mtok out$0.042per Mtok cachedai-gateway.vercel.sh · read 2026-09-20 |
| ElectronHub aggregator | standard | $0.85per Mtok in$2.5per Mtok outapi.electronhub.ai · read 2026-11-01 |
| Venice AI aggregator | standard | $1.04per Mtok in$3.13per Mtok out$0.052per Mtok cachedmodels.dev · read 2026-11-01 |
| PPIO aggregator | standard | $6per Mtok in$18per Mtok outapi.ppinfra.com · read 2026-11-01 |
Measured 1 standing
| Place | Board | Metric | Score |
|---|---|---|---|
| 16thof 84 | Proofbench — Epoch AI | Accuracy | 75 |
About
Hy4-preview — a text model from Tencent Hunyuan, sold by 11 companies from $0.42 in and $1.25 out per million tokens, placed 16th of 84 on Proofbench — Epoch AI.
It takes text and returns text, with a context window of 1,048,576 tokens. It was published in August 2026. Its sellers say it can reason step by step and call a tool. The catalogue files it under chat. Its weights are published under APACHE-2.0, so you may run it on your own machine, or buy it from whoever serves it cheapest. Eleven companies sell it. The cheapest is $0.42 in and $1.25 out per million tokens at Nous Research. Beside the standard rate there are separately routed and regional lanes. It stands 16th of 84 on Proofbench — Epoch AI.
Every current figure
- Maker
- Tencent Hunyuan
- Register
- model
- Takes
- text
- Returns
- text
- Context
- 1,048,576 tokens
- Longest answer
- 65,536 tokens
- Published
- August 2026
- Parameters
- 780 billion · read from its own weights
- Licence
- apache-2.0
- Sellers
- 11
- Maker's own price
- not read
- Price
- $0.42 in and $1.25 out per million tokens — Nous Research
- Boards
- 1
- Best place
- 16th of 84 — Proofbench — Epoch AI
Known as 1 name
tencent/Hy4-preview