
Hy4 flash liegt bei 0,070 US-Dollar pro Million Eingabe-Token: Tencents günstigstes Hy-Modell ist zugleich das am schlechtesten dokumentierte.
- GoogleNEUGoogle: Gemini 4 Carbon2026-11-10$2.00 / $10.00 pro 1 Mio. Tokens · 90 tok/s
- AnthropicNEUAnthropic: Claude Fable 5.22026-11-09$8.00 / $40.00 pro 1 Mio. Tokens · 97 tok/s
- orcaNEUOrca: OrcaC2 1.02026-11-0717 tok/s
- OrcaNEUOrcaCyber: GLM 5.3 Abliterated / Uncensored2026-11-04$3.00 / $5.00 pro 1 Mio. Tokens · 87 tok/s
- OrcaNEUOrcaCyber: Hy4 Abliterated / Uncensored2026-11-04$0.80 / $1.60 pro 1 Mio. Tokens · 133 tok/s
- OrcaNEUOrcaCyber: Kimi K3.1 Pro Abliterated / Uncensored2026-11-04$4.00 / $20.00 pro 1 Mio. Tokens · 72 tok/s
- OrcaOrcaCyber: MiniMax M3.1 Flash Abliterated / Uncensored2026-10-29$0.80 / $1.60 pro 1 Mio. Tokens · 130 tok/s
- OrcaOrcaCyber: GLM 5.3 Flash Abliterated / Uncensored2026-10-27$0.80 / $1.60 pro 1 Mio. Tokens · 65 tok/s
- OrcaOrcaCyber: Qwen3.8 Flash Next Abliterated / Uncensored2026-10-26$0.80 / $1.60 pro 1 Mio. Tokens · 92 tok/s
- OrcaOrcaCyber: MiMo V2.6 Flash Abliterated / Uncensored2026-10-26$0.80 / $1.60 pro 1 Mio. Tokens · 43 tok/s
- MoonshotMoonshotAI: Kimi K3.1 Pro2026-10-2447Intelligenz
- OrcaOrcaCyber: DeepSeek V4.1 Flash Abliterated / Uncensored2026-10-23$0.80 / $1.60 pro 1 Mio. Tokens · 193 tok/s
- minimaxMiniMax: M3.1 Flash2026-10-1929Intelligenz
- OrcaOrca: OrcaCyber Ultra 1.02026-10-17$5.00 / $10.00 pro 1 Mio. Tokens · 150 tok/s
- deepseekDeepSeek: DeepSeek V4.1 Flash 10172026-10-1733Intelligenz
- DeepSeekDeepSeek: DeepSeek V4.1 Flash2026-10-1740Intelligenz
- OrcaOrca: OrcaCyber Zero 1.52026-10-10$3.00 / $7.50 pro 1 Mio. Tokens · 119 tok/s
- openaiOpenAI: GPT-6.1 Sol2026-09-2955Intelligenz
- anthropicAnthropic: Claude Sonnet 5.52026-09-2855Intelligenz
- typesafeTypeSafe: Jev 1.132026-09-24$0.04 / $0.00 pro 1 Mio. Tokens · 219 tok/s
Tencent released Hy4 flash on 16 November 2026, and the specification reads like a cost-engineering exercise more than a capability one: 224 billion total parameters of which 64 billion are Engram parameters, 12 billion active per token, native multimodality, a 1,048,576-token context, and a mainland list price of ¥0.5 per million input tokens against ¥2 per million output. The company's framing is that it matches Claude Haiku 5.5 on everyday work in a blind test run on its own employees, at roughly an eighteenth of the per-task cost. The blind test is real and the price is real; the comparison across them is Tencent's own arithmetic, and the difference between those two statements is most of what anyone deploying this model needs to know.
There is no model card. There is no architecture paper, no licence, no weights repository and no independent benchmark run. What Tencent published instead is a research page with benchmark results baked into PNG charts, a set of internal product testimonials, and a promise to open-source the model later. That is a thinner dossier than Hy4 preview shipped with in August, and the gap matters more here than it would for a flagship, because the entire pitch for Hy4 flash is efficiency — and efficiency claims are the ones you cannot check from a price table.
Was die Preisliste tatsächlich aussagt
Per million tokens, on the mainland card: ¥0.5 input, ¥2 output, ¥0.02 cached input. Tencent quotes the same model in dollars as $0.070 input, $0.281 output and $0.003 cached, which is the figure the launch post led with. Set it beside the rest of the family and the positioning is unambiguous.
• Hy4 flash — ¥0,5 Eingabe / ¥2 Ausgabe / ¥0,02 zwischengespeichert, 224B insgesamt mit 12B aktiv.
• Tencent HY4 — ¥6 input / ¥18 output / ¥0.3 cached, 770B total with 49B active, the general-availability build from 4 November 2026.
• Hy4-Vorschau — derselbe Tarif von ¥6 / ¥18 / ¥0,3, Apache 2.0, Modellgewichte seit 28. August 2026 herunterladbar.
Das ist eine zwölffache Senkung bei der Eingabe und eine neunfache bei der Ausgabe gegenüber dem eigenen Flaggschiff, von einer Generation zur nächsten, ohne Änderung am Kontextfenster. Die zwischengespeicherte Eingabe fällt um das Fünfzehnfache – genau der Posten, der die Wirtschaftlichkeit jedes Agenten entscheidet, der bei jedem Schritt einen langen System-Prompt erneut sendet. Für eine Arbeitslast, die ein großes Dokument einmal liest und dann zwanzig Fragen dazu stellt, ist die Cache-Rate die entscheidende Rate, und 0,003 $ pro Million sind so bepreist, dass dieses Muster nahezu kostenlos wird.
Two promotions sit on top of the list price and both are temporary. Tencent Cloud's TokenHub is running a 50% discount for a limited period with no stated end date, and WorkBuddy is free to use until 30 November. Neither should be modelled into a production budget; the list rate is the number that survives December.
Woher 12 Milliarden aktive Parameter kommen

The parameter split is the first thing to understand, because 64 billion Engram parameters inside a 224-billion-parameter model is not a normal arrangement. Engram, in Tencent's usage, is a memory-style component that holds learned associations rather than participating in every forward pass the way an expert does. The practical consequence is that a large share of the model's total size is not charged against per-token compute: 12 billion parameters activate per token, which is a quarter of what GLM-5.3 Flash activates and roughly a quarter of Tencent HY4's 49 billion.
Tencent names four levers, and each one attacks a different part of the bill.
Pre-training. Hybrid attention combined with LatentMoE and the Engram parameters. Tencent's stated result is a KV cache one twenty-fourth the size of Hy3's and one sixth of Hy4's. On long-context serving that is the difference between a request fitting on a node and a request needing a cluster, and it is the single most consequential number in the release for anyone renting GPUs by the hour.
Post-training. Self-distillation from Tencent HY4, then reinforcement learning. The stated outcome is a 50% reduction in end-to-end output length. This is where a flash model usually gives back part of its price advantage — shorter answers can mean shallower work — and it is also where the per-task cost claims below get most of their leverage. Output tokens are the expensive ones at $0.281 per million, so halving them halves the larger side of the invoice before the price cut is even applied.
Inference. A megakernel, four-bit quantization and a decoding path Tencent calls DFly, together sustaining better than 150 tokens per second of decode. Fast decode changes agent behaviour as much as it changes user experience: a tool-calling loop that spends less wall-clock per step can afford more steps inside the same latency budget.
Architekturdetail. Nichts veröffentlicht. Layer-Anzahl, Expert-Anzahl, Routing-Breite, die Größe des Vision-Encoders und ob Multimodalität nativ oder nachträglich angebaut ist, fehlen alle. Die Zahlen 224B und 12B stammen von der Forschungsseite; alles feiner Aufgeschlüsselte existiert öffentlich nicht.
Der Blindtest und was „matches Claude Haiku 5.5“ abdeckt
The claim that travelled is the blind test. Tencent ran it with 85 users of its WorkBuddy product across 27 task categories and 259 real daily-work tasks, comparing Hy4 flash with several unnamed alternatives and reporting that user satisfaction came out close to Claude Haiku 5.5, that Hy4 flash had the lowest per-task cost of any model in the panel, and that it finished tasks in the shortest time.
Lies die Formulierung sorgfältig, denn der Launch-Beitrag und die Forschungsseite sagen nicht dasselbe. Der Beitrag sagt, das Modell „entspreche“ Claude Haiku 5.5. Die Seite sagt, die Zufriedenheit sei nahe daran gewesen. Nahe ist nicht gleich, und ein Anbieter, der sein eigenes Panel zusammenfasst, wird die Formulierung wählen, die schmeichelt. Noch wichtiger: Das Panel besteht aus 85 Mitarbeitenden des Unternehmens, das das Modell entwickelt hat, die an Aufgaben arbeiten, die das Unternehmen ausgewählt hat, in einer Oberfläche, die das Unternehmen ebenfalls entwickelt. Das ist eine vernünftige Art zu entscheiden, ob ein Modell intern ausgeliefert werden sollte. Es ist kein Beleg dafür, dass die Erfahrungen Dritter übereinstimmen werden, und es belegt keine Gleichwertigkeit bei Aufgabenarten, die in den 27 Kategorien zufällig nicht enthalten waren.
Was der Test tatsächlich belegt, ist enger gefasst und dennoch nützlich: Bei büroähnlicher Arbeit – Dokumente, Präsentationen, Tabellen, also genau das, wofür WorkBuddy gedacht ist – war ein Modell mit 12B aktiven Parametern nach Einschätzung von Personen, die beide täglich nutzen, nicht offensichtlich schlechter als ein Wettbewerber nahe der Frontier. Wenn sich das verallgemeinern lässt, ist das Kostenargument überwältigend. Ob es sich verallgemeinern lässt, ist die offene Frage, und es gibt keinen öffentlichen Benchmark in der Veröffentlichung, der sie beantworten würde.
Nachverfolgung der 18x
The "18x task cost reduction" in the launch post is not a price ratio. Hy4 flash costs twelve times less per input token than Tencent HY4, so a simple per-token comparison gives 12x, not 18x. The extra comes from Tencent's WorkIndex, an internal eight-suite benchmark covering PowerPoint creation, PowerPoint revision, human-machine co-writing, Markdown co-writing, document editing, table repair, audio and video processing, and visual understanding. On that suite, at the maximum tier and at busy-hour prices, Tencent reports per-task cost at one-thirteenth of Hy4 preview, one-twentieth of Tencent HY4, and one-eighth of DeepSeek-V4.1-Flash-1017.
One-twentieth against its own flagship and one-eighteenth against the model the post names is the same order of magnitude, so the 18x is a defensible round number — but it is a per-task figure, which means it folds in output length, retry rate and step count alongside price. Two of those three inputs are things Tencent controls through post-training and reports only as a percentage. A model that answers in half the tokens at a twelfth of the price is not 18x cheaper because of the price; it is 18x cheaper because it is also being made to talk less. If your task needs the longer answer, you get less than 18x, and the article you are reading cannot tell you how much less.
The same caveat applies to every per-task figure in the release, including the internal product numbers Tencent volunteered: a 70% cut in task time and roughly one-fifteenth the cost in WorkBuddy against Tencent HY4, a 2.52% general-chat gain against GLM-5.3-Flash in 元宝, 29% fewer output tokens and 6% fewer steps in 微信支付 agent flows against DeepSeek-V4.1-Flash-1017, 88% faster output and 60% fewer thinking tokens in 企业微信, and 22% to 39% fewer requests with 40% to 61% fewer tokens in 王者荣耀 root-cause analysis. These are the most credible numbers in the package, because they come from production systems with real users and real bills. They are also unaudited and selectively released, and no product publishes the flows where the model lost.
Was das Release nicht enthält
No open weights. Tencent says it plans to open-source Hy4 flash and to iterate on it frequently. A plan is not a licence. Hy4 preview shipped under Apache 2.0 on day one with a Hugging Face repository, vLLM and SGLang recipes and published sampling defaults; Hy4 flash has none of that, and the wording around the open-source commitment is softer than the preview's was.
No independent evaluation. As of publication, Artificial Analysis has not posted a score for Hy4 flash, and no third party has reproduced a single number from the release. Every figure above is Tencent's.
Keine maschinenlesbare Benchmark-Tabelle. Die Ergebnisse von DeepSWE, Terminal-Bench 4.0, APEX-Agents, AutomationBench, HLE und OSWorld 2.1 liegen auf der Anbieterseite als Diagrammbilder vor, nicht als Text. Das ist eine seltsame Wahl für einen Launch, dessen ganzes Argument messbare Effizienz ist, und es bedeutet, dass die Zahlen nicht kopiert, per Diff verglichen oder überprüft werden können, ohne Pixel zu lesen.
Keine Sicherheits- oder Systemkarte. Auch keine Aussage zum Ablehnungsverhalten, zur Datenverarbeitung oder zur regionalen Verfügbarkeit über die genannten Festlandprodukte hinaus.
Wo es läuft und wie man es günstig ausprobieren kann
Availability on launch day is Tencent's own surface: TokenHub on Tencent Cloud for API access, WorkBuddy for office tasks free until 30 November, and 元宝, 企业微信, QQ, ima and Marvis on the consumer and enterprise side. Nothing else is announced.

On OrcaRouter, Hy4 flash is not in the catalogue yet — the route that exists for this family is tencent/hy4-preview, at $0.834 per million input and $2.501 per million output, alongside a rate-limited free tier. That is the model Hy4 flash was distilled from, so it is the right baseline to measure against while you wait: run the same tasks on both once the flash route lands and you will have the per-task ratio that Tencent's WorkIndex implies without having to trust it.
The reason to care about where it is served is the price model. OrcaRouter passes provider list price through at 0% markup, which is not a discount but a commitment about what happens when a vendor cuts one: Tencent has already run two promotional discounts on this model in its first week, and a pass-through route reflects the next one the day it appears rather than when a reseller updates a table. Failover matters for the same reason in reverse — an unproven 12B-active model on launch-week infrastructure is exactly the model you want to be able to fall back from without rewriting a client, and the same key reaches more than 200 models if it does.
Worauf man es zuerst richten sollte
Tencent's own deployment guidance splits the family cleanly, which is more useful than most vendor advice because it concedes a weakness: Hy4 flash for deep search, document processing, PowerPoint generation, simple frontend work and computer use; Tencent HY4 for long-horizon software engineering. The concession is in the last clause — the cheap model is not being sold as a coding replacement for the expensive one, and the internal WorkBuddy data shows a small drop on long coding tasks even as it reaches parity on daily work.

Das lässt sich auf einen einfachen Testplan abbilden. Nimm zwanzig Aufgaben, die du bereits ausführst, von denen die meisten dokument- oder agentenlastig sind und einige wirklich schwierige Engineering-Aufgaben. Erfasse Eingabe-Tokens, Ausgabe-Tokens, Cache-Treffer, Wall-Clock-Zeit und ob die Ausgabe brauchbar war. Lass sie auf Hy4 flash und auf der Baseline laufen, die du heute verwendest. Du wirst an einem Nachmittag erfahren, ob die 18x auf deinen Traffic zutrifft – mehr, als dir das Release selbst sagen wird.
Das günstigste Hy-Modell ist das am schlechtesten lesbare.
Tencent has built something worth paying attention to: a quarter-active model with a million-token window, a KV cache small enough to change the serving maths, and a price that undercuts its own flagship by an order of magnitude. The engineering described — Engram parameters, a 50% output-length cut through self-distillation, four-bit inference at over 150 tokens per second — is the kind of work that shows up in a bill rather than a leaderboard, which is presumably why Tencent led with per-task cost.
But the release asks to be believed on exactly the quantities it declines to document. No weights, no model card, no independent score, benchmark numbers behind pixels, and a headline comparison drawn from a panel of its own employees. None of that is dishonest. All of it means the honest summary is "this is probably the best value in the Hy family and nobody outside Tencent has verified it," and the right response is a two-week trial with token counts, not a migration.
The open-source promise is the thing to diary. If Tencent ships weights the way it did for Hy4 preview — Apache 2.0, day one, with serving recipes — the efficiency claims become checkable and the model becomes a serious option for anyone who wants to run cheap inference themselves. Until that happens, Hy4 flash is a very good price attached to a very thin dossier, and the difference between those two facts is where your evaluation budget should go.
In diesem Artikel verglichen3
Aus diesem Artikel erkannt · Benchmarks: Artificial Analysis · täglich aktualisiert
