Tencent · Available now
Hy4 Flash by Tencent (2026): price per reply, price per million tokens and context window
Hy4 Flash is a lower-cost model in Tencent’s Hy family, announced on November 16, 2026 with 224B total and 12B active parameters, a context of up to 1M tokens and a Tencent Cloud list price of $0.07 per million input and $0.281 per million output tokens (settled at half price until November 30, 2026).
Use it now
Chat
≈ 1 credit per reply · See all models
New accounts get 5 free credits once, valid for 7 days. They work on the lower-priced models here; the priciest ones need credits you bought.
Start chatting — 5 free credits- Maker
- Tencent
- Type
- Text
- Released
- 2026-11-16
- OpenRouter id
- tencent/hy4-flash
Per million tokens
Price
| - | Input | Output | Source |
|---|---|---|---|
| Tencent — List price (TokenHub) | $0.07 | $0.28 | tencentcloud.com |
| Via OpenRouter (tencent/hy4-flash) | $0.04 | $0.14 | openrouter.ai |
Checked
Facts, with sources
- Tencent’s Hy4 flash announcement, dated November 16, 2026, presents it as the newest member of the Hy model family. Tencent recommends it for most everyday tasks (deep research, document processing, PowerPoint generation, simple front-end work, computer use) and points complex tasks such as long-horizon software engineering to Hy4. Tencent Cloud’s API guide calls it "a leaner, more efficient everyday workhorse".Source: tencent.com · checked 2026-11-17
- The Tencent announcement is dated November 16, 2026, and Tencent Cloud’s model, pricing and API pages for hy4-flash were last updated the same day. The announcement says only that Hy4 flash is available now; it gives no separate release date or version name.Source: tencent.com · checked 2026-11-17
- Tencent says Hy4 flash has 224B total parameters (including 64B Engram parameters), 12B active parameters, a context of up to 1M tokens and multimodal input; the announcement does not state a maximum output or the output types. It says the KV cache is 1/6 the size of Hy4’s and that decoding exceeds 150 tokens per second. These are the maker’s own figures, not tested by us.Source: tencent.com · checked 2026-11-17
- Tencent Cloud’s model list gives hy4-flash a 1M-token context window, 960k maximum input and 64k maximum output, with deep reasoning (preserved thinking), structured output, function calling, caching and image understanding. The page was last updated November 16, 2026.Source: tencentcloud.com · checked 2026-11-17
- Tencent Cloud’s model pricing page lists hy4-flash at $0.07 per million input tokens, $0.281 per million output tokens and $0.003 per million cache-hit tokens, with no context-length tiers (hy4 on the same page: $0.834 / $2.501 / $0.042). It says hy4-flash is settled at 50% of the list price until 12:00:00 Beijing time on November 30, 2026, after which the list price applies. Tencent’s announcement gives the same list prices as "approximately" $0.070 / $0.281 / $0.003, so the two Tencent pages agree.Source: tencentcloud.com · checked 2026-11-17
- Tencent says Hy4 flash is available on Tencent Cloud TokenHub, WorkBuddy, Yuanbao, WeCom, QQ, ima and Marvis. It says WorkBuddy use is free until November 30, 2026 and TokenHub is at 50% of the standard price. The announcement says OpenRouter integration is "coming soon"; OpenRouter’s own listing was already live when we checked on November 17, 2026 (see the OpenRouter fact).Source: tencent.com · checked 2026-11-17
- Tencent Cloud’s API guide lists model name hy4-flash on TokenHub (base URL tokenhub-intl.tencentcloudmaas.com/v1), compatible with the OpenAI Chat Completions, OpenAI Responses and Anthropic Messages APIs. Thinking is turned on with "thinking": {"type": "enabled"}, and reasoning_effort defaults to high for hy4-flash.Source: tencentcloud.com · checked 2026-11-17
- OpenRouter lists tencent/hy4-flash with Tencent as the only endpoint: 1,048,576-token context, 65,536-token maximum output, text and image input and text output, $0.035 / $0.1405 per million input / output tokens with $0.0015 cache read and a 0.5 discount field (half of Tencent’s list price), fp8 quantization, and support for reasoning, tools and structured outputs. OpenRouter describes it as "a cost-efficient variant of Hy4" with roughly a quarter of the flagship’s parameter count (checked November 17, 2026). Unlike tencent/hy4, its page lists no reasoning-effort levels.Source: openrouter.ai · checked 2026-11-17 · second-hand
- Tencent’s Hugging Face page for tencent/Hy4-flash is public and not gated (created November 17, 2026), and its card says Hy4 flash, Hy4 flash-FP8 and the pre-trained Hy4 flash-base are released as open weights on Hugging Face, ModelScope and GitCode. The card says 220B total parameters (including 64B Engram) and 12B active; the announcement says 224B. The licence is the Tencent Hy License, which is derived from Apache 2.0 but is not Apache 2.0: a business that gives others remote access to the model (MaaS) needs a separate written licence from Tencent if its worldwide revenue is over US$50 million a year or the product has over 100 million monthly active users.Source: huggingface.co · checked 2026-11-20
- Tencent’s model card links a GGUF version, AngelSlim/Hy4-flash-GGUF, which it says is for “end-to-end testing purposes”. The repository lists five builds, from 125.73 GiB (IQ2-AH) to 156.79 GiB (Q4_K_M, called the safe default), so the files alone are over 125 GiB. Other uploaders have posted their own GGUF files; we did not check those and have not run any of them.Source: huggingface.co · checked 2026-11-20
Hy4 family
Hy4 Flash next to Hy4 and Hy4 Preview
- Side by side, Hy4 Flash has 224B total and 12B active parameters and a Tencent Cloud list price of $0.07 / $0.281 per million input / output tokens, against $0.834 / $2.501 for Hy4; OpenRouter shows $0.035 / $0.1405 for Hy4 Flash and $0.417 / $1.2505 for Hy4 (checked November 18, 2026).Source: openrouter.ai · checked 2026-11-18
Side by side
Hy4 Flash, Hy4 and Hy4 Preview compared
People who search for Hy4 Flash usually want to know how it differs from the larger Hy4. By Tencent’s numbers it is the smaller and cheaper one: 12B active parameters, and a list price roughly a tenth of Hy4’s. Tencent also settles Hy4 Flash at half the list price until November 30, 2026, which is why OpenRouter shows $0.035 / $0.1405. You can chat with Hy4 Flash on this page; a typical reply costs about 1 credit, as it does for the Qwen3.8 Flash model.
| What | Hy4 Flash | Hy4 | Hy4 Preview |
|---|---|---|---|
| Announced by Tencent | November 16, 2026 | November 4, 2026 | August 28, 2026 |
| Total / active parameters (Tencent) | 224B / 12B | Not stated by Tencent | 770B / 49B |
| Tencent Cloud list price per million tokens (input / output) | $0.07 / $0.281 | $0.834 / $2.501 | About $0.90 / $2.70 |
| OpenRouter price per million tokens (input / output) | $0.035 / $0.1405 | $0.417 / $1.2505 | $0.834 / $2.501 |
| Input | Text and image | Text and image | Text |
| Context window / max output (OpenRouter) | 1,048,576 / 65,536 tokens | 1,048,576 / 65,536 tokens | 1,048,576 / 64,000 tokens |
| Typical reply at the OpenRouter price (our arithmetic) | about 1 credit | about 1 credit | about 1 credit |
Source: tencent.comSource: tencent.comSource: tencent.comSource: tencentcloud.comSource: openrouter.ai
Nearest available
Models you can use now instead
Tencent Hy4
Tencent$0.42 / $1.2505 (per million tokens)
- Tencent’s Hy4 announcement, dated November 4, 2026, calls Hy4 its "latest-generation large language model". Compared with Hy4 Preview it adds visual understanding (a first for the Hy line) and computer use, and Tencent says it improves instruction following, long-context comprehension and long-horizon agent tasks. The maker’s own figures, not tested by us: hallucination rate in Tencent’s internal tests fell from 12.95% to 7.2%.Source: tencent.com · checked 2026-11-17
- Tencent published the Hy4 launch announcement on November 4, 2026, as the full release that follows Hy4 Preview. The announcement does not state a parameter count or a context length for Hy4.Source: tencent.com · checked 2026-11-17
- Hy4 Preview, the earlier version, was announced by Tencent on August 28, 2026 as open source, with 770B total and 49B active parameters and a context window "exceeding 1M tokens". Tencent’s GitHub page for it shows an Apache 2.0 licence badge. These are the preview’s figures; we found no Tencent page that gives the same numbers for Hy4.Source: github.com · checked 2026-11-17
DeepSeek V4.1 Flash
DeepSeek$0.30 / $1.2 (per million tokens)
- DeepSeek released V4.1-Flash on September 10, 2026.Source: api-docs.deepseek.com · checked 2026-11-11
- Only the Flash version of V4.1 exists: DeepSeek has published no V4.1 Pro, and the Pro model on its price page is still deepseek-v4-pro (V4-Pro-0813).Source: api-docs.deepseek.com · checked 2026-11-11
- DeepSeek’s pricing page says the API model deepseek-flash is now served by the V4.1-Flash-1017 snapshot.Source: api-docs.deepseek.com · checked 2026-11-11
FAQ
Questions
What is Hy4 Flash?
Hy4 Flash is a model from Tencent, announced on November 16, 2026 as the newest member of its Hy family. Tencent says it has 224B total parameters (including 64B Engram parameters) and 12B active, takes multimodal input and suits everyday tasks such as deep research, document processing, PowerPoint generation, simple front-end work and computer use. OpenRouter calls it "a cost-efficient variant of Hy4". It is a separate model from Hy4, not a setting of it.
How much does Hy4 Flash cost?
Tencent Cloud’s pricing page lists $0.07 per million input tokens, $0.281 per million output tokens and $0.003 per million cache-hit tokens, settled at 50% of list until 12:00 Beijing time on November 30, 2026. OpenRouter lists tencent/hy4-flash at $0.035 / $0.1405, which is half of the list price (checked November 17, 2026); we do not know what OpenRouter will charge once Tencent’s half-price period ends. On this site a typical reply to this model costs about 1 credit (a credit is about $0.01 in the $5 pack), billed by its real token cost, at least 1 credit.
How do I use Hy4 Flash?
Tencent says it is on Tencent Cloud TokenHub (API model name hy4-flash, compatible with the OpenAI Chat Completions, OpenAI Responses and Anthropic Messages APIs), WorkBuddy, Yuanbao, WeCom, QQ, ima and Marvis. It is also on OpenRouter as tencent/hy4-flash. You can also chat with it in the chat box on this page, on credits.
What is the context window of Hy4 Flash?
Tencent says up to 1M tokens. Tencent Cloud’s model list gives a 1M-token context window with up to 960k tokens of input and up to 64k tokens of output. OpenRouter shows 1,048,576 tokens of context and a 65,536-token maximum output.
Is Hy4 Flash open source, and is there a GGUF version (hy4 flash gguf)?
Its weights are open, under Tencent’s own licence rather than a standard open-source one. Tencent’s Hugging Face card (tencent/Hy4-flash) says it released Hy4 flash, Hy4 flash-FP8 and the pre-trained Hy4 flash-base as open weights on Hugging Face, ModelScope and GitCode (checked November 20, 2026). The Tencent Hy License is derived from Apache 2.0 but is not Apache 2.0: a business that sells remote access to the model needs a separate written licence from Tencent if its worldwide revenue is over US$50 million a year or the product has over 100 million monthly active users. The card also links a GGUF version, AngelSlim/Hy4-flash-GGUF, made for end-to-end testing: five builds from 125.73 GiB to 156.79 GiB, so even the smallest file is over 125 GiB; we have not run them. The chat on this page uses the hosted model, not a download.
How is Hy4 Flash different from Hy4?
Tencent recommends Hy4 Flash for most everyday tasks and Hy4 for complex ones such as long-horizon software engineering. On Tencent Cloud, Hy4 Flash lists at $0.07 / $0.281 per million tokens against $0.834 / $2.501 for Hy4. Tencent says Hy4 Flash’s KV cache is 1/6 the size of Hy4’s; that is the maker’s own claim, not tested by us. Tencent’s announcement does not give Hy4’s parameter count; OpenRouter says Hy4 Flash has roughly a quarter of the flagship’s.
Is this the official Tencent website?
No. This is an unofficial page, not affiliated with or endorsed by Tencent. Facts link to Tencent’s own pages where possible.