
Hy4 flash arriveert op $0.070 per miljoen input: het goedkoopste Hy-model van Tencent is ook het minst gedocumenteerde.
- GoogleNIEUWGoogle: Gemini 4 Carbon2026-11-10$2.00 / $10.00 per 1 mln tokens · 90 tok/s
- AnthropicNIEUWAnthropic: Claude Fable 5.22026-11-09$8.00 / $40.00 per 1 mln tokens · 97 tok/s
- orcaNIEUWOrca: OrcaC2 1.02026-11-0717 tok/s
- OrcaNIEUWOrcaCyber: GLM 5.3 Abliterated / Uncensored2026-11-04$3.00 / $5.00 per 1 mln tokens · 87 tok/s
- OrcaNIEUWOrcaCyber: Hy4 Abliterated / Uncensored2026-11-04$0.80 / $1.60 per 1 mln tokens · 133 tok/s
- OrcaNIEUWOrcaCyber: Kimi K3.1 Pro Abliterated / Uncensored2026-11-04$4.00 / $20.00 per 1 mln tokens · 72 tok/s
- OrcaOrcaCyber: MiniMax M3.1 Flash Abliterated / Uncensored2026-10-29$0.80 / $1.60 per 1 mln tokens · 130 tok/s
- OrcaOrcaCyber: GLM 5.3 Flash Abliterated / Uncensored2026-10-27$0.80 / $1.60 per 1 mln tokens · 65 tok/s
- OrcaOrcaCyber: Qwen3.8 Flash Next Abliterated / Uncensored2026-10-26$0.80 / $1.60 per 1 mln tokens · 92 tok/s
- OrcaOrcaCyber: MiMo V2.6 Flash Abliterated / Uncensored2026-10-26$0.80 / $1.60 per 1 mln tokens · 43 tok/s
- MoonshotMoonshotAI: Kimi K3.1 Pro2026-10-2447Intelligentie
- OrcaOrcaCyber: DeepSeek V4.1 Flash Abliterated / Uncensored2026-10-23$0.80 / $1.60 per 1 mln tokens · 193 tok/s
- minimaxMiniMax: M3.1 Flash2026-10-1929Intelligentie
- OrcaOrca: OrcaCyber Ultra 1.02026-10-17$5.00 / $10.00 per 1 mln tokens · 150 tok/s
- deepseekDeepSeek: DeepSeek V4.1 Flash 10172026-10-1733Intelligentie
- DeepSeekDeepSeek: DeepSeek V4.1 Flash2026-10-1740Intelligentie
- OrcaOrca: OrcaCyber Zero 1.52026-10-10$3.00 / $7.50 per 1 mln tokens · 119 tok/s
- openaiOpenAI: GPT-6.1 Sol2026-09-2955Intelligentie
- anthropicAnthropic: Claude Sonnet 5.52026-09-2855Intelligentie
- typesafeTypeSafe: Jev 1.132026-09-24$0.04 / $0.00 per 1 mln 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.
Wat de tariefkaart eigenlijk zegt
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 invoer / ¥2 uitvoer / ¥0,02 gecachet, 224B totaal met 12B actief.
• Tencent HY4 — ¥6 input / ¥18 output / ¥0.3 cached, 770B total with 49B active, the general-availability build from 4 November 2026.
• Hy4 preview — hetzelfde tarief van ¥6 / ¥18 / ¥0,3, Apache 2.0, gewichten downloadbaar sinds 28 augustus 2026.
Dat is een twaalfvoudige verlaging van de input en een negenvoudige verlaging van de output ten opzichte van zijn eigen flagship, van de ene generatie op de andere, zonder wijziging aan het contextvenster. Gecachte input daalt met factor vijftien, en dat is de kostenpost die de economie bepaalt van elke agent die bij elke stap een lange systeemprompt opnieuw verzendt. Voor een workload die een groot document één keer leest en er daarna twintig vragen over stelt, is het cachetarief het tarief dat telt, en $0.003 per miljoen is geprijsd om dat patroon vrijwel gratis te maken.
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.
Waar 12 miljard actieve parameters vandaan komen

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.
Architectuurdetails. Niets gepubliceerd. Het aantal lagen, het aantal experts, de routeringsbreedte, de grootte van de vision-encoder en of multimodaliteit native is of er later aan vastgeschroefd is: het ontbreekt allemaal. De cijfers 224B en 12B komen van de onderzoekspagina; alles wat fijner is, bestaat niet in het openbaar.
De blindtest, en wat "overeenkomt met Claude Haiku 5.5" omvat
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.
Lees de formulering aandachtig, want het lanceringsbericht en de onderzoekspagina zeggen niet hetzelfde. Het bericht zegt dat het model Claude Haiku 5.5 "evenaart". De pagina zegt dat de tevredenheid dicht bij die van het model lag. Dichtbij is niet gelijk, en een leverancier die zijn eigen panel samenvat, kiest de framing die gunstig overkomt. Belangrijker: het panel bestaat uit 85 werknemers van het bedrijf dat het model heeft gebouwd, die werken aan taken die het bedrijf heeft geselecteerd, in een interface die het bedrijf ook zelf bouwt. Dat is een redelijke manier om te beslissen of je een model intern uitrolt. Het is geen bewijs dat de resultaten bij een derde partij zullen overeenkomen, en het stelt geen gelijkwaardigheid vast voor taaktypen die de 27 categorieën toevallig niet omvatten.
Wat de test wel aantoont, is beperkter en toch nuttig: bij werk dat op kantoorwerk lijkt — documenten, slides, spreadsheets, precies het soort werk waarvoor WorkBuddy bedoeld is — was een 12B-actief model niet duidelijk slechter dan een concurrent die dicht bij de frontier staat, afgaand op het oordeel van mensen die beide dagelijks gebruiken. Als dat generaliseert, is het kostenargument overweldigend. Of het generaliseert, is de open vraag, en er zit geen publieke benchmark in de release die daar antwoord op zou geven.
De 18x traceren
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.
Wat de release niet bevat
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.
Geen machineleesbare benchmarktabel. De resultaten voor DeepSWE, Terminal-Bench 4.0, APEX-Agents, AutomationBench, HLE en OSWorld 2.1 bestaan als grafiekafbeeldingen op de leverancierspagina, niet als tekst. Dat is een vreemde keuze voor een lancering waarvan het hele argument meetbare efficiëntie is, en het betekent dat de cijfers niet kunnen worden gekopieerd, vergeleken of gecontroleerd zonder pixels te lezen.
Geen veiligheids- of systeemkaart. Ook geen verklaring over weigeringsgedrag, gegevensverwerking of regionale beschikbaarheid buiten de genoemde producten voor het vasteland.
Waar het draait, en hoe je het goedkoop kunt uitproberen
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.
Waar je het eerst op moet richten
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.

Dat sluit mooi aan bij een simpel testplan. Neem twintig taken die je al uitvoert, waarvan de meeste document- of agentachtig zijn en een paar echt moeilijke engineeringtaken. Registreer inputtokens, outputtokens, cachehits, wall-clocktijd en of de output bruikbaar was. Draai ze op Hy4 flash en op de baseline die je vandaag gebruikt. Je leert in één middag of de 18x van toepassing is op jouw verkeer, wat meer is dan de release zelf je zal vertellen.
Het goedkoopste Hy-model is het minst leesbare
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.
Vergeleken in dit artikel3
Herkend uit dit artikel · Benchmarks: Artificial Analysis · dagelijks bijgewerkt
