Hunyuan Models

9 modelsGeneral models free to startUp to 1.05M context

Usage

Last 30 days · 2026-10-21 to 2026-11-19

Tokens

4.1B

Requests

27.7K

Models in use

8 of 9

Tokens per day, stacked by model

0357M713M10-2110-2811-0411-1111-182026-10-21 — 162,497,680 tokens hy4-preview: 161,918,605 hy3: 579,0752026-10-22 — 619,998,730 tokens hy4-preview: 600,929,615 hy3: 19,069,1152026-10-23 — 48,475,545 tokens hy4-preview: 46,271,070 hy3: 2,204,4752026-10-24 — 9,232,105 tokens hy4-preview: 9,028,875 hy3: 203,2302026-10-25 — 524,785 tokens hy4-preview: 383,185 hy3: 141,6002026-10-26 — 1,277,305 tokens hy3: 1,184,030 hy4-preview: 93,2752026-10-27 — 15,244,405 tokens hy3: 9,557,960 hy4-preview: 5,686,4452026-10-28 — 59,685 tokens hy3: 55,435 hy4-preview: 4,2502026-10-29 — 35,070,885 tokens hy4-preview: 34,728,685 hy3: 342,2002026-10-30 — 59,465 tokens hy4-preview: 43,565 hy3: 15,9002026-10-31 — 7,114,510 tokens hy3: 6,966,985 hy4-preview: 147,5252026-11-01 — 131,324,970 tokens hy3: 131,321,455 hy4-preview: 3,5152026-11-02 — 1,390,755 tokens hy3: 1,375,160 hy4-preview: 15,5952026-11-03 — 23,730 tokens hy3: 18,275 hy4-preview: 5,4552026-11-04 — 444,209,965 tokens hy4: 433,960,625 hy-3d-3.1: 7,500,230 hy4-preview: 2,222,195 hy3: 526,9152026-11-05 — 14,954,370 tokens hy4: 14,290,105 hy3: 516,230 hy4-preview: 148,0352026-11-06 — 118,126,025 tokens hy4: 117,166,755 hy-3d-3.1: 833,250 hy4-preview: 102,480 hy3: 23,5402026-11-07 — 733,160 tokens hy4: 581,880 hy3: 147,235 hy4-preview: 4,0452026-11-08 — 2,838,500 tokens hy4-preview: 2,106,505 hy-3d-3.1: 625,000 hy3: 58,675 hy4: 48,3202026-11-09 — 5,387,505 tokens hy4-preview: 4,368,230 hy-3d-3.1: 833,250 hy4: 144,445 hy3: 41,5802026-11-10 — 72,837,760 tokens hy4-preview: 67,057,265 hy4: 4,021,855 hy-3d-3.1: 1,666,745 hy3: 91,8952026-11-11 — 37,043,595 tokens hy4-preview: 28,206,235 hy4: 8,626,610 hy3: 210,7502026-11-12 — 81,492,445 tokens hy4-preview: 44,164,545 hy4: 37,221,210 hy3: 106,6902026-11-13 — 116,722,715 tokens hy4: 116,337,945 hy3: 380,700 hy4-preview: 4,0702026-11-14 — 7,305,675 tokens hy3: 5,018,735 hy4: 2,283,480 hy4-preview: 3,4602026-11-15 — 13,546,565 tokens hy4: 12,273,460 hy3: 879,860 hy4-preview: 393,2452026-11-16 — 340,676,335 tokens hy4-flash: 336,744,825 hy4: 2,847,340 hy3: 943,150 hy4-preview: 141,0202026-11-17 — 712,367,305 tokens hy4-flash: 551,210,775 hy4-preview: 121,898,935 hy4: 38,092,755 hy3: 1,164,8402026-11-18 — 390,288,710 tokens hy4: 155,932,890 hy3: 129,305,145 hy4-flash: 104,361,410 hy4-preview: 689,2652026-11-19 — 713,348,530 tokens hy4-flash: 693,071,940 hy4: 15,751,260 hy3: 2,587,510 hy4-preview: 1,937,820
  • hy4-flash
  • hy4-preview
  • hy4
  • hy3
  • hy-3d-3.1

Which models that traffic went to

  1. Hy4 Flash41.1%1.7B
  2. Hy4 Preview27.6%1.1B
  3. Hy423.4%960M
  4. Hy37.7%315M
  5. Hy 3d 3.10.3%11.5M

Share of 4.1B tokens. 3 models with traffic report no token counts and cannot be ranked here, including hy3-free and hy3-preview — they are in the request view.

The two views disagree on purpose: a model can take a large share of the calls and a small share of the tokens — many short requests — or the reverse. Which one matters depends on whether your cost is driven by call volume or by prompt length. Measured on AIHubMix over the last 30 days, counting the 9 model IDs listed on this page; traffic routed through upstream-specific IDs that are not in the public catalog is not included.

All 9 Hunyuan Models

Open in model list
Hunyuan models on AIHubMix with input and output modalities, context length, maximum output, price per million tokens including cache read and cache write rates, and measured throughput and latency.
Modalities
hy4-previewTakes text, returns text.1.05M64K$0.845$2.535/M$0.0423/M19 tok/s3.52 s
hy3-freeTakes text, returns text.256K128KFreeFree/M———
hy3Takes text, returns text.256K128K$0.1562$0.6248/M$0.0391/M43 tok/s2.38 s
hy3-previewTakes text, returns text.256K128K$0.17$0.5667/M$0.051/M40 tok/s2.05 s
hy4-flashTakes text, vision, returns text.——$0.081$0.324/M$0.0032/M60 tok/s3.46 s
tencent/Hunyuan-A13B-InstructTakes , returns text.——$0.14$0.56/M———
tencent/Hunyuan-MT-7BTakes , returns text.——$0.2$0.2/M—8 tok/s17.93 s
hy4Takes text, vision, returns text.——$0.8823$2.647/M$0.4412/M78 tok/s8.65 s
hy-3d-3.1Takes text, vision. Output modality not published.——See pricing———

Prices are USD per million tokens; cache read and cache write are the rates for prompt-cache hits and for writing a prompt into the cache. Throughput and latency are measured on AIHubMix — the same figures the model detail page shows — not vendor claims. A dash means the catalog does not publish that field for that model, which is not the same as the model not supporting it.

Hunyuan on AIHubMix

Which Hunyuan model should I start with?

hy3-free is free on input — the cheapest entry here that declares tool calling, and it carries a 256K context. Move up to hy4 when answer quality matters more than cost, or to hy4-preview for long-form reasoning.

Which of these models reason before answering?

4 of the 9 models here declare a reasoning phase — they work through the problem before producing an answer, which helps on multi-step problems at the cost of extra output tokens. Use the Reasoning filter above the table to see them. The catalog does not record anything further about how they differ, so this page does not sort them into families.

Why are there several entries for the same model?

Because each row is a route you can call, not a model release. Some IDs name an upstream (azure-, alicloud-, cc-), some are the open-weight repository form (tencent/…), and some differ only in capitalisation, kept so older integrations keep working.

The catalog does not carry a field saying which of those a given row is, so this page does not sort them into buckets it would have to invent. Every row shows that route’s own price, context and speed — compare those directly, and open a model to see the upstreams that serve it.

How is cached input billed?

The Cache read column is the rate for input tokens served from the prompt cache — for example hy4-flash bills cache hits at 4% of the input rate and hy4-preview bills cache hits at 5% of the input rate. Cache write is the surcharge for putting a prompt into the cache in the first place, and only a few upstreams bill it separately. A dash in either column means the catalog carries no cache rate for that model, so plan on paying the full input rate.

Do I need a separate Hunyuan account?

No. One AIHubMix key covers every model on this page, and switching between them is a change to the model string — billing, rate limits, and logs stay in one place.

Start calling Hunyuan in one line

One key, one endpoint, 983 models across 49 model authors.