· CostMyTokens · Analysis · 6 min read
Hunyuan Hy4 API Pricing: $0.417/$1.2505 Promo Through Dec 31, 2026
Hunyuan Hy4 API pricing: TokenHub promo $0.417/$1.2505/$0.021 through Dec 31, 2026; list $0.834/$2.501/$0.042. Official Tencent USD rates vs DeepSeek, GLM and Kimi.
Quick answer: Tencent’s Hunyuan Hy4 (API id hy4) officially lists at $0.834 per 1M input tokens, $2.501 per 1M output tokens and $0.042 per 1M cached input tokens on Tencent Cloud’s international TokenHub price table. Through December 31, 2026 (23:59:59 Beijing time), TokenHub is billing Hy4 inference at 50% of list, so the rate you actually pay today is $0.417 input / $1.2505 output / $0.021 cached input per 1M tokens. On the standard workload used across this site (10M tokens a month, 8M input / 2M output), that is $5.84 a month at the promo rate and $11.67 at list. Prices in US dollars, from Tencent’s official international pricing page, verified November 5, 2026.
Disclosure: this article contains no affiliate links — Tencent has no cash affiliate program, and the Hy4 pages on this site carry none. Where CostMyTokens does use affiliate links elsewhere, they never affect order; every table is sorted by price only. Read our disclosure.
Hy4 official prices: promo rate vs list rate
| Rate | Input / 1M | Output / 1M | Cached input / 1M |
|---|---|---|---|
| TokenHub promo (through Dec 31, 2026) | $0.417 | $1.2505 | $0.021 |
| Official list (from Jan 1, 2027) | $0.834 | $2.501 | $0.042 |
Three things to know before you budget around these numbers:
- The promo is a dated discount, not the price. Like Gemini 4 Argon’s introductory pricing, Hy4’s half rate has a published end date. Anything you build on $0.417/$1.2505 costs exactly double from January 1, 2027, unless TokenHub extends the offer. The Hy4 model page calculator uses the promo rate while it runs and shows the list rate alongside it.
- Cached input is about 5% of the input rate at both promo and list ($0.021 against $0.417), which makes Hy4 unusually cheap for cache-heavy agent loops and long-context reuse. There is no published cache-write charge in the international table.
- There is no published USD batch rate for Hy4. The mainland batch table lists hy4-preview only, and no matching batch row appears in the international documentation, so we show Batch as not published rather than estimating one. The full model list and rate provenance are on our Tencent Hunyuan provider page.
A note on the mainland CNY catalog: Tencent publishes a separate mainland China price catalog for Hy4 at CNY 6 input / CNY 18 output / CNY 0.3 cached input per 1M tokens. That is a different regional price schedule, not a currency conversion of the USD table — the two do not reconcile at any single exchange rate, which is why this site never converts CNY rows into USD figures. If you buy through mainland Tencent Cloud, the CNY catalog is your price; if you buy through the international site, the USD table above is.
What Hy4 costs in practice
At the current promo rate (worked examples from the Hy4 model page, verified November 5, 2026):
| Workload | Per call | Per month |
|---|---|---|
| Chat assistant — 2,000 in / 500 out per call, 10,000 calls/month | $0.001459 | $14.59 |
| Document processing — 50,000 in / 2,000 out per call, 1,000 calls/month | $0.0234 | $23.35 |
| Agent loop with caching — 10,000 fresh in / 100,000 cached in / 2,000 out per call, 500 calls/month | $0.008771 | $4.39 |
For smaller units: 1000 tokens cost $0.000417 as input and $0.0012505 as output at the promo rate — see our explainer on how much 1000 tokens cost in dollars. Run your own volumes through the cost calculator or the preset calculator on the Hy4 model page.
Hy4 vs DeepSeek, GLM and other models at its price point
Official first-party rates only, sorted by the standard workload (10M tokens/month, 8M input / 2M output). Time-of-day and promo rows are labelled — compare like with like.
| Model | Input / 1M | Output / 1M | Standard workload / month |
|---|---|---|---|
| GLM-5.3-Flash | $0.15 | $0.50 | $2.20 |
| DeepSeek V4.1 Flash (off-peak) | $0.15 | $0.60 | $2.40 |
| DeepSeek V4.1 Flash (peak) | $0.30 | $1.20 | $4.80 |
| GLM-5.3-FlashX | $0.37 | $1.25 | $5.46 |
| Hunyuan Hy4 (promo, to Dec 31, 2026) | $0.417 | $1.2505 | $5.84 |
| DeepSeek V4 Pro (off-peak) | $0.66 | $1.98 | $9.24 |
| Hunyuan Hy4 (list, from Jan 1, 2027) | $0.834 | $2.501 | $11.67 |
| Kimi K2.6 (legacy / closing) | $0.95 | $4.00 | $15.60 |
| DeepSeek V4 Pro (peak) | $1.32 | $3.96 | $18.48 |
| GLM-5.3 | $1.40 | $4.40 | $20.00 |
* Kimi K2.6 is listed as legacy on this site (closing schedule); keep it for historical comparison only. Prefer current Kimi K3 / K3.1 Pro rates on our Kimi API pricing page.
The honest reading of that table: at the promo rate Hy4 undercuts every flagship-tier rival in its class — it is less than a third of GLM-5.3’s standard-workload cost and beats even DeepSeek V4 Pro’s off-peak rate. What it does not beat is the Flash tier: GLM-5.3-Flash, DeepSeek V4.1 Flash and GLM-5.3-FlashX are all cheaper per token, at both promo and list. And the promo expiry matters for ranking: at list price from January, Hy4 lands mid-pack — still cheaper than Kimi K2.6, DeepSeek V4 Pro at peak and GLM-5.3, but no longer cheaper than DeepSeek V4 Pro off-peak. Full rate cards: DeepSeek API pricing, GLM-5.3 API pricing, and the full ranking in our cheapest LLM API comparison.
Hy4 vs Hy4 Preview
Hy4 Preview (hy4-preview) carries the same TokenHub list rates as Hy4 — $0.834 / $2.501 / $0.042 — so there is no price reason to pick one over the other; choose on capability and stability. Hy4 is the general-availability flagship released November 4, 2026, with a 1M-token context window; the Preview is the earlier checkpoint whose weights Tencent released openly. Per-model detail: Hy4 and Hy4 Preview.
FAQ
How much does the Hunyuan Hy4 API cost? The official international list price is $0.834 per 1M input tokens, $2.501 per 1M output tokens and $0.042 per 1M cached input tokens. Through December 31, 2026, TokenHub bills at 50% of list: $0.417 / $1.2505 / $0.021.
When does the Hy4 50% discount end? December 31, 2026 at 23:59:59 Beijing time. From January 1, 2027 the list rates apply unless Tencent extends the promotion.
Why do Chinese sources quote Hy4 at ¥6 and ¥18? Mainland Tencent Cloud publishes a separate CNY catalog (¥6 input / ¥18 output / ¥0.3 cached input per 1M tokens). It is a regional price schedule, not a conversion of the USD table, and the two should not be mixed.
Is Hy4 cheaper than DeepSeek? Against DeepSeek’s flagship V4 Pro, yes at the promo rate ($5.84 vs $9.24–$18.48 on the standard workload), and at list price it still beats V4 Pro’s peak rate. Against the DeepSeek V4.1 Flash tier, no — Flash is cheaper at every Hy4 rate.
Does Hy4 have a batch discount? No published USD batch rate exists in Tencent’s international documentation (the mainland batch table lists hy4-preview only), so we do not quote one.
Is Hy4 the same price as Hy4 Preview? At list, yes — both are $0.834 / $2.501 / $0.042 per 1M tokens on TokenHub. The 50% promotional billing through December 31, 2026 is published for Hy4 inference.
Related pages
Price Hy4 on the Hy4 model page or Hy4 Preview, browse the full lineup on the Tencent Hunyuan provider page, and run your own volumes on the cost calculator. Compare with DeepSeek API pricing, GLM-5.3 API pricing, the cheapest LLM API ranking, and how much 1000 tokens cost in dollars.
Prices verified against Tencent Cloud’s official international TokenHub pricing documentation on November 5, 2026, and cross-checked with the Hy4 model and Tencent provider pages tracked on this site. Tencent states international prices in USD per million tokens and mainland prices in CNY; the two catalogs are published separately and are never converted into each other here.