A generated title card reading 'Hy4 flash vs GLM-5.3-Flash' with the subtitle 'The cheaper model is the one you cannot download', a price line reading '$0.070 / $0.281 / $0.003 against $0.15 / $0.50 / $0.03 per million tokens', and a footer reading 'Hy4 flash released 16 November 2026; GLM-5.3-Flash open-sourced 26 August 2026.'
Guides & Insights

Hy4 flash vs GLM-5.3-Flash: The Cheaper Model Is the One You Cannot Download

Author

jinhao.song

Date Published

Latest models · 20View all models →
Benchmarks: Artificial Analysis · updated daily
Back to all posts

Hy4 flash and GLM-5.3-Flash are aimed at the same buyer — someone who wants a million-token context window, image input and agent-grade tool calling without paying flagship rates — and they answer with almost identical price cards. Hy4 flash, released by Tencent on 16 November 2026, bills $0.070 per million input tokens and $0.281 per million output with $0.003 cached. GLM-5.3-Flash, which Zhipu launched and open-sourced on 26 August 2026, bills $0.15, $0.50 and $0.03 at list. Tencent is roughly half the price on input and a little over half on output, and ten times cheaper on cached input, which is the line that decides most agent budgets. It is also the model with no weights, no licence, no model card and no independent benchmark, against one that is MIT-licensed on Hugging Face and scores 27.8 on the Artificial Analysis Intelligence Index.

That is the whole comparison in one sentence, and it does not resolve cleanly. Cheaper and unverifiable against pricier and inspectable is a real trade, not a marketing problem, and which side wins depends on things neither vendor will tell you. What follows is what each has actually published, what the price difference buys and costs, and where the decision turns.

Two price cards, read line by line

Start with the numbers as published, because they are the only figures in this comparison that both companies state plainly and that anyone can check against a bill.

• Input, per million tokens — Hy4 flash $0.070; GLM-5.3-Flash $0.15.

• Output, per million tokens — Hy4 flash $0.281; GLM-5.3-Flash $0.50.

• Cached input, per million tokens — Hy4 flash $0.003; GLM-5.3-Flash $0.03.

• Context window — Hy4 flash 1,048,576 tokens; GLM-5.3-Flash 1,000,000 tokens.

• Active parameters per token — Hy4 flash 12B of 224B total, including 64B of Engram parameters; GLM-5.3-Flash 18B of 320B total across 45 layers.

• Weights — Hy4 flash announced as a plan with no date; GLM-5.3-Flash MIT on Hugging Face since launch day.

• Independent intelligence score — Hy4 flash none published; GLM-5.3-Flash 27.8 on the Artificial Analysis Intelligence Index, with a 71.5 on the same lab's Coding Index.

The cached-input line deserves more attention than it usually gets. At $0.003 per million, Hy4 flash caches cheaper than GLM-5.3-Flash by a factor of ten, and cheaper than its own flagship Tencent HY4 by a factor of a hundred. For an agent that re-sends a fixed system prompt, a tool schema and a retrieved document on every step, the cache rate is the rate you actually pay most of the time, and a tenfold difference there outweighs the twofold difference on fresh input. If your workload is retrieval-heavy, Hy4 flash's advantage is larger than the headline price suggests. If your workload is one-shot generation with no reuse, the advantage narrows to the input and output ratios.

Both price cards are promotional in part. Tencent Cloud's TokenHub is running 50% off for an unstated limited period and WorkBuddy is free until 30 November. Zhipu's 50%-off launch rate expired on 9 September 2026 and was not extended, which is why the $0.075 and $0.25 figures that circulated in late August no longer apply — the rate above is the standing one. Assume Tencent's promotion ends the same way and budget the list rate.

The parameter gap is smaller than the price gap

A generated comparison card titled 'Two cheap models, two designs' with a left column for Hy4 flash reading 224B total, 12B active, 64B Engram parameters, 1M context, $0.070 / $0.281 / $0.003, no weights published, and a right column for GLM-5.3-Flash reading 320B total, 18B active, 45 layers, 1M context, $0.15 / $0.50 / $0.03, MIT weights on Hugging Face. A footer reads 'Hy4 flash figures are Tencent's, published 16 November 2026; GLM-5.3-Flash price and licence per Zhipu, 26 August 2026.'

Hy4 flash activates 12 billion parameters per token against GLM-5.3-Flash's 18 billion, a third less compute per token, and that is the mechanical reason it can be sold at half the price. The interesting part is the total: 224 billion against 320 billion, with 64 billion of Tencent's total sitting in Engram parameters that hold learned associations rather than running in every forward pass the way a routed expert does. Tencent has not published a layer count, expert count or routing width for Hy4 flash, so the closest thing to an architectural comparison is that both are sparse mixture-of-experts models with a small active fraction and both vendors chose that design for the same reason — serving cost scales with active parameters, not total ones.

GLM-5.3-Flash's design is documented in more detail, and that documentation is checkable because the weights are public: 45 layers, hybrid sparse-plus-linear attention, an approach to residual connections Zhipu calls Manifold-Constrained Hyper-Connections, and an IndexPool mechanism that compresses four indexer key vectors into one weighted pool to cut long-context latency. Pre-training ran on a 30-trillion-token multimodal corpus. None of that is independently audited, but it is at least a description specific enough to be falsified by anyone who downloads the checkpoint.

The asymmetry is not that Zhipu published and Tencent did not. Both published; they published different kinds of thing. Zhipu described a model, which is a claim about structure that a checkpoint can confirm or refute. Tencent described a result — a KV cache a sixth the size of its flagship's, output length halved by self-distillation, better than 150 tokens per second of decode from four-bit quantization and a megakernel — which is a claim about behaviour that only reproduction can settle. Results are the more commercially interesting claims and the harder ones to check. On a model with public weights, a serving claim gets tested by strangers within weeks; on a model without them, it stays a sentence on a research page for as long as the vendor wants.

One of these you can download

This is the deciding difference for a meaningful share of buyers, and it is not close. GLM-5.3-Flash is MIT-licensed, in FP8 and BF16, at zai-org/GLM-5.3-Flash on Hugging Face, and has been since 26 August 2026. MIT permits commercial use, modification, fine-tuning and redistribution with no revenue threshold. The model has been in wide public use for nearly three months, which means the inference stack has met it: quantizations exist, serving recipes exist, and any claim about its behaviour can be tested by running it.

Nothing like it exists for Hy4 flash. The commitment there is a sentence about future plans, hedged with an intention to keep iterating, and it names no licence and no date. Tencent has done this properly before — the August preview landed under Apache 2.0 with weights on four hosting sites and vLLM and SGLang recipes on day one — so the pattern is not in doubt. What is in doubt is the timidity of the current language, which reads like a commitment to a process rather than to an artefact. Anyone planning to fine-tune, self-host, or build a cost model around running their own inference should treat Hy4 flash as API-only until a repository appears.

The same asymmetry applies to evidence. GLM-5.3-Flash has two numbers from a harness neither vendor wrote: 27.8 on the Artificial Analysis Intelligence Index and 71.5 on its Coding Index. Both are checkable, and the pair says something specific: a 71.5 on coding against 27.8 on the general index describes a model that is genuinely good at one thing and ordinary at everything else, which is exactly what a cheap tier should be. That is the kind of distinction a single headline score erases, and it is why the remaining vendor figures in this piece are presented as claims rather than results. Hy4 flash has no independent score, and its benchmark results — DeepSWE, Terminal-Bench 4.0, APEX-Agents, OSWorld 2.1 and Tencent's internal suites — exist only as chart images on the vendor's research page. That is not a reason to distrust them. It is a reason to treat them as hypotheses.

What each vendor measured, and who it beat

Both companies published comparisons, and they compared themselves to different things, which tells you what each thought its weakness was.

Tencent's headline is a blind test: 85 WorkBuddy users, 27 task categories, 259 real daily-work tasks, with Hy4 flash reported as landing close to Claude Haiku 5.5 on satisfaction while carrying the lowest per-task cost and shortest task duration in the panel. Note "close to," not "better than," and note the panel is the vendor's own employees in the vendor's own product. The supporting claim is per-task cost on WorkIndex, Tencent's internal eight-suite benchmark: one-thirteenth of Hy4 preview, one-twentieth of Tencent HY4, one-eighth of DeepSeek-V4.1-Flash-1017. Against GLM-5.3-Flash the only published figure is a 2.52% gain in general chat inside 元宝, which is a product metric on one consumer app and does not transfer to anything you would run.

Zhipu's positioning is different: it sells the model as frontier-adjacent capability at flash cost, prices it against its own 743B flagship, and points to open weights and an exploit-evaluation result as proof the cheap model is not a toy. Its internal figures have already taken one independent correction, and the AA number is the one to quote.

The practical consequence is that neither vendor has published a head-to-head on a neutral harness, and the comparison you would actually want — same tasks, same prompts, same harness, both models — does not exist. You will have to run it. Ten tasks you already do, token counts recorded, is enough to settle it for your traffic and is worth more than either launch page.

Two different bets on multimodality

GLM-5.3-Flash takes text, image and video input natively and was the first GLM-5 model to do so rather than routing vision through a separate adapter. That is a documented, testable claim, and the weights are public so it can be checked without asking permission.

Hy4 flash is described as multimodal, and the WorkIndex suite it was measured on includes visual understanding as one of eight task types. Tencent has not said whether the vision path is native or an adapter, has not published the encoder's size, and has not said whether video is supported at all. The most that can be said is that image input is part of the pitch and that the model's predecessor in the family — Tencent HY4, general availability 4 November 2026 — added native vision and computer use as its headline feature, so the capability is plausible for a model distilled from it. Plausible is not documented.

If screenshots, charts, scanned documents or video frames are load-bearing in your pipeline, GLM-5.3-Flash is the model whose multimodal behaviour you can verify before committing. If images are occasional, the gap matters less than the price.

Where the decision turns

Choose Hy4 flash when the volume is the problem and the task is office-shaped. Document processing, retrieval over a long corpus, slide generation, simple frontend work, computer-use loops — these are the workloads Tencent built it for, the ones its internal product data covers, and the ones where a tenfold cache discount and a 50% output-length cut compound into something an order of magnitude cheaper. Run it on your own tasks first and watch output tokens: if the model really is terser, your bill falls further than the rate card predicts, and if it is terser because it is shallower, you will see that in the same measurement.

Choose GLM-5.3-Flash when you need the artefact, the audit, or the modality. Fine-tuning, on-premises serving, regulated data, a licence you can show a lawyer, video input, or a score from a harness nobody at Zhipu wrote — all of these point one way. At $0.15 and $0.50 it is not expensive; it is roughly a tenth of GLM-5.3's list price for a model that has already survived independent measurement.

Choose both when the workload is mixed. The honest reading of the two releases is that they are not really substitutes: one is a cost-optimised API model for high-volume everyday work, the other is an open checkpoint with documented architecture for anything that needs to leave the vendor's control. A router that reaches both is a cheaper answer than picking one and being wrong.

A screenshot of the OrcaRouter model page for GLM 5.3 Flash, slug z-ai/glm-5.3-flash, credited to Z.ai and dated 2026-08-26, describing a 320B-total, 18B-active native multimodal model with a 1M-token context that takes text, image and video in and returns text. The specification strip reads max output 128K, input $0.15 and output $0.50 per million tokens, p50 time to first token 3.82 seconds and p95 10.00 seconds. The pricing panel lists $0.150 input, $0.500 output and $0.030 cache read per million tokens, and the public benchmark panel shows 71.5 on the AA Coding index and 27.8 on AA Intelligence. A button reads 'Get the GLM 5.3 Flash API'.

What is actually callable today

This is where the two models differ operationally as well as commercially. GLM-5.3-Flash is in the OrcaRouter catalogue as z-ai/glm-5.3-flash at the provider's list price of $0.15 per million input and $0.50 per million output, with a rate-limited free tier alongside it for evaluation. Hy4 flash is not routed here — Tencent has not made it available to us — so anyone comparing the two needs Tencent's own TokenHub for the flash side and can use the same OrcaRouter key for the GLM side.

That split is worth using rather than resenting. OrcaRouter passes provider list price through with no markup, which means the GLM side of your comparison is billed at exactly Zhipu's rate and stays correct if Zhipu moves it; there is no reseller table to go stale. And for the evaluation itself, having tencent/hy4-preview on the same key as z-ai/glm-5.3-flash is useful even though it is not the model in question: the preview is the model Hy4 flash was distilled from, so it is the natural control when you test whether the small model's answers hold up. One key, more than 200 models, and a failover path if a launch-week endpoint turns out to be unreliable — which, for a model released six days ago with no public uptime record, is a reasonable thing to want.

A generated decision card titled 'Pick by constraint' with three rows: high volume office-shaped work points to Hy4 flash on price per task; fine-tuning, on-premises serving or a named licence points to GLM-5.3-Flash under MIT; mixed workloads point to routing both from one key. A footer reads 'GLM-5.3-Flash scores 27.8 on the Artificial Analysis Intelligence Index; Hy4 flash has none published.'

The price is not the question

Tencent has built the cheaper model, and on the numbers it is cheaper by a margin that matters: half on input, a little over half on output, a tenth on cache. What it has not built is a model you can inspect. No weights, no licence, no architecture detail, no independent score, benchmarks behind pixels, and a headline comparison drawn from a panel of its own staff. Zhipu built the more expensive one and published enough that its weaknesses — including a vendor index that independent measurement cut by fifteen points — are on the record where you can weigh them.

Six days after launch, that is the shape of the choice: a demonstrated discount against an undocumented model, or a documented model at a price that is still very low. The discount is probably real and the documentation gap is probably temporary. Neither is certain, and the way to find out is to spend an afternoon measuring rather than a quarter choosing.

Compared in this article3

Detected from this article · Benchmarks: Artificial Analysis · updated daily