Tencent · llm model
Hy4 flash
Hy4 flash is Tencent's latest language model in our dataset, released Nov 16, 2026. Open weights are promised but not yet published. Tencent's Nov 16, 2026 cost-focused Hy model: 224 billion total parameters (including 64 billion Engram parameters) with 12 billion active, multimodal inputs and a 1-million-token context, on Tencent Cloud TokenHub, WorkBuddy and other Tencent apps; Tencent plans to open-source the weights later. Verified against vendor sources on Nov 17, 2026.
Reported Last verified · All models · JSON · Markdown
Primary source: https://www.tencent.com/tencent-launches-hy4-flash-to-power-high-speed-cost-efficient-ai-deployment-for-everyday-workflows/. We are the fast, independent index; the vendor’s own page is the source of record for these model facts.
Hy4 flash at a glance: 1,000,000-token context window; listed price $0.07 in / $0.281 out per 1M tokens.
- ReleasedNov 16, 2026
- Context window1M1,000,000 tokens
- LicenceWeights unpublishedNot published
Modalities not verified
Key facts
Vendor-reported facts are read from the linked vendor page on the date shown above; we did not run the model. Independently tested means we ran it; no fact on this page is. Editorial is our own classification, not a vendor claim. Where a linked page does not itself state a value, the row says so. How to read provenance.
| Fact | Value | Label | Source |
|---|---|---|---|
| Vendor | Tencent | Editorial | n/a |
| Released | Nov 16, 2026 Tencent's newsroom announcement is dated November 16, 2026. | Vendor-reported | source |
| Status | generally available | Editorial | n/a |
| Context window | 1,000,000 tokens (1M) | Vendor-reported | source |
| Parameters | 224B total, 12B active, hybrid attention, LatentMoE, and Engram Tencent says the 224 billion total includes 64 billion Engram parameters, with 12 billion active. | Vendor-reported | source |
| Licence | Not published Tencent says it plans to open-source Hy4 flash with frequent iterations; no weights or license were published as of Nov 17, 2026. | Vendor-reported | source |
| Pricing | $0.07 in / $0.281 out per 1M tokens Tencent's listed standard API prices, written as approximately US$0.070, US$0.281 and US$0.003 per million tokens. A 50 percent TokenHub discount through Nov 30, 2026 is described on the same page and is not applied here. Prices: tencent.com/tencent-launches-hy4-flash-to-power-high-speed-cost-efficient-ai-deployment-for-everyday-workflows · | Vendor-reported | source |
Where Hy4 flash sits in our data
- Released Nov 16, 2026 (week 47 of 2026), in November 2026. Other models in our data released that month: Qwen4-Max, LongCat-Video-Avatar 1.6, Gemini 4 Hydrogen, Gemini Decision 1.
- The previous Tencent model we track is Hy4 (Nov 4, 2026), 12 days earlier.
- Of the 77 models in our data with a published context window, Hy4 flash has the number 23 largest (1M tokens). Our data is not every model on the market.
- All releases by date: AI model releases by month · new AI models this week. Choosing for code: best LLM for coding. Price and rank context: LLM pricing comparison · cheapest LLM API · OpenRouter pricing · best AI model right now.
Alternatives to Hy4 flash
Language models from other vendors that share at least one of our groups with Hy4 flash. Comparable on paper, from each model’s sourced facts; we have not run them.
Side by side: Longest context window LLMs
How Hy4 flash differs from Hy4
Read from the recorded facts of the two models, not from a head-to-head test. We have not run either.
- Listed price is $0.07 in and $0.281 out, against $0.834 in and $2.501 out for Hy4 (per 1M tokens).
- Both list a 1M-token context window.
- Hy4 flash came out 12 days after Hy4.
- Facts for the other model: Hy4.
What the sources say
These are claims from the linked pages, not our measurements.
- Vendor-reported (Tencent), not verified: Hy4 flash slightly outperforms Hy4 Preview on Tencent's internal WorkIndex, at a lower per-task cost; a blind evaluation with 85 internal WorkBuddy users across 259 tasks is said to have shown satisfaction close to Claude Haiku 5.5, with the lowest per-task cost and the shortest completion time among the models they evaluated. We did not verify these evaluations.
- Vendor-reported (Tencent), not verified: reinforcement learning cut average end-to-end task output tokens by 50 percent, and inference optimizations are said to push decoding speed past 150 tokens per second. We did not measure this.
Notes
- Tencent recommends Hy4 flash for everyday tasks such as document processing, slide generation and computer use, and Hy4 for complex long-horizon software engineering.
- The announcement also points to Tencent's research post at https://hy.tencent.ai/research/hy4-flash for a price comparison table.
What we tested
Nothing yet. We have not run Hy4 flash ourselves, so every fact on this page is Reported from the sources below. When we do test it, the scope will be stated here and in our test logs. See methodology.
Not verified, left out
We could not confirm these on a vendor page, so we do not state them.
- Tencent says it plans to open-source Hy4 flash; we did not find published weights, a license or a system card as of Nov 17, 2026.
- Tencent's announcement does not give an API model id. It says OpenRouter integration is coming soon.
- From Nov 16 through Nov 30, 2026, Tencent says Hy4 flash is free on WorkBuddy and 50 percent off standard TokenHub API pricing. The prices below are the standard list prices in the announcement, not the discounted TokenHub rate.
Hy4 flash FAQ
When was Hy4 flash released?
Tencent released Hy4 flash on Nov 16, 2026. Source.
How much does Hy4 flash cost?
The listed API price is $0.07 in / $0.281 out per 1M tokens. Prices change; the source shows the current figure. Source.
How large is the Hy4 flash context window?
1,000,000 tokens. Source.
Is Hy4 flash open weights?
No. The weights are not published; the licence is listed as Not published. Source.
Has Agentic Weekly tested Hy4 flash?
Not yet. Every fact on this page is Reported: read from the linked vendor pages, not measured by us.
Sources
Compare and guides
We have no head-to-head page or guide that names Hy4 flash yet. These hubs cover the models we do compare.
All model comparisons · All guides · Model releases by month
More from Tencent
- Hy4 (Nov 4, 2026)
Newsletter
The Weekly Verdict. Free, every Monday.
A free email every Monday: the week's few big AI releases, then 5 short use cases for the new models, each with a prompt you can paste.
Free. Double opt-in. One click to unsubscribe.