Hunyuan Hy3 vs Hunyuan Hy4 Preview

Tencent · China  |  Tencent · China · Updated June 2026

Quick verdict

Both are Tencent models. Hunyuan Hy4 Preview is the newer, generally stronger default; reach for Hunyuan Hy3 when its lower price or a specific cost or latency profile matters more than the latest capabilities.

Hunyuan Hy3 and Hunyuan Hy4 Preview are both Tencent models, so the real question is not which lab to trust but which tier fits your workload and budget. Hunyuan Hy3 is a 295B Apache-2.0 open MoE that reaches frontier reasoning quality while running at roughly 21B active-parameter cost. Hunyuan Hy4 Preview is tencent's open-sourced (Apache 2.0) next-generation Hunyuan flagship preview, released August 28, 2026 as a 770B-parameter (49B active) MoE model that helped optimize its own training pipeline. Since both come from the same lab, the comparison below focuses on the tier-and-cost trade-offs that actually separate them.

Key differences at a glance

Side-by-side specs

SpecHunyuan Hy3Hunyuan Hy4 Preview
ProviderTencent (China) Tencent (China)
ReleasedJuly 6, 2026 August 28, 2026
Context window256K (~384 pages) 1M+ tokens (~1,500 pages)
Price (in/out)Open weight (self-host / free) $0.834/$2.501 per 1M tokens
Open weight?Yes — self-hostable Yes — self-hostable
Modalitiestext, code text
SWE-Bench VerifiedNot published Not published
MRCR v2 @ 1MNot published Not published

Who wins what

Frontier-level reported reasoning and science (GPQA Diamond 90.4) at low active-parameter cost

Hunyuan Hy3

Hunyuan Hy3 lists frontier-level reported reasoning and science (GPQA Diamond 90.4) at low active-parameter cost among its strengths; Hunyuan Hy4 Preview does not.

Runs a 295B model at the cost of a 21B — only 21B parameters active per token

Hunyuan Hy3

Hunyuan Hy3 lists runs a 295B model at the cost of a 21B — only 21B parameters active per token among its strengths; Hunyuan Hy4 Preview does not.

Clean, unrestricted Apache-2.0 license with no geographic carve-out

Hunyuan Hy3

Hunyuan Hy3 lists clean, unrestricted Apache-2.0 license with no geographic carve-out among its strengths; Hunyuan Hy4 Preview does not.

GPQA Diamond (92.3)

Hunyuan Hy4 Preview

Tencent's open-sourced (Apache 2.0) next-generation Hunyuan flagship preview, released August 28, 2026 as a 770B-parameter (49B active) MoE model that helped optimize its own training pipeline — and it carries the larger 1M+ tokens context.

Terminal-Bench (85.4)

Hunyuan Hy4 Preview

Hunyuan Hy3 is comparatively weak here — benchmarks are largely self-reported, and the ultra-low hosted pricing is a limited promotion

SWE-bench Multilingual (82.9)

Hunyuan Hy4 Preview

Tencent's open-sourced (Apache 2.0) next-generation Hunyuan flagship preview, released August 28, 2026 as a 770B-parameter (49B active) MoE model that helped optimize its own training pipeline — and it is the newer of the two.

Lowest cost at scale

Hunyuan Hy3

Its weights are open, so at volume you pay for your own hardware instead of Hunyuan Hy4 Preview's $0.834/$2.501 per 1M tokens.

Largest single-prompt input

Hunyuan Hy4 Preview

Its 1M+ tokens window is about 3.9× larger than Hunyuan Hy3's 256K, fitting roughly 1,500 pages in one prompt.

Which should you pick?

A cost-sensitive startup shipping high volume

Hunyuan Hy3

At Open weight (self-host / free) it undercuts Hunyuan Hy4 Preview, and on millions of tokens that margin decides the monthly bill.

Someone analysing very long documents or codebases

Hunyuan Hy4 Preview

Larger 1M+ tokens window fits more in one prompt.

Anyone whose priority is frontier-level reported reasoning and science (gpqa diamond 90.4) at low active-parameter cost

Hunyuan Hy3

It is specifically built for that.

Anyone whose priority is gpqa diamond (92.3)

Hunyuan Hy4 Preview

That is its strongest area.

Hunyuan Hy3: where it fits

A 295B Apache-2.0 open MoE that reaches frontier reasoning quality while running at roughly 21B active-parameter cost. Released July 6, 2026 by Tencent, it is built for frontier-level reported reasoning and science (GPQA Diamond 90.4) at low active-parameter cost, runs a 295B model at the cost of a 21B — only 21B parameters active per token, clean, unrestricted Apache-2.0 license with no geographic carve-out, and broad day-one ecosystem support plus an FP8 checkpoint.

Its trade-offs are real: benchmarks are largely self-reported, and the ultra-low hosted pricing is a limited promotion, and the hosted API is China-jurisdiction, and self-hosting a 295B MoE still needs serious hardware. As an open-weight model, its running cost is your own hardware rather than a per-token fee.

Hunyuan Hy4 Preview: where it fits

Tencent's open-sourced (Apache 2.0) next-generation Hunyuan flagship preview, released August 28, 2026 as a 770B-parameter (49B active) MoE model that helped optimize its own training pipeline. Released August 28, 2026 by Tencent, it is built for gPQA Diamond (92.3), terminal-Bench (85.4), sWE-bench Multilingual (82.9), and large jump in long-horizon coding vs prior Hunyuan model (DeepSWE 28.0 to 64.3).

Its trade-offs: text-only, no native vision support, labeled a 'preview', not yet Tencent's finalized GA flagship, and sWE-Marathon score still low in absolute terms (31.9) despite a large relative jump. At $0.834 in / $2.501 out per million tokens, it sits in the budget price band.

The bottom line for this matchup

Because Hunyuan Hy3 and Hunyuan Hy4 Preview come from the same lab (Tencent), they share the same training philosophy and ecosystem — the decision is purely tier vs. cost. Hunyuan Hy4 Preview is the more capable, more recent option; the other earns its place only when its price or latency profile fits a specific job better. Most teams should default to Hunyuan Hy4 Preview and drop down only with a concrete reason.

Want both Hunyuan Hy3 and Hunyuan Hy4 Preview without two subscriptions? LumiChats gives you these plus 40+ models under one ₹69/day pass (about $1/day) — draft with one, cross-check with the other.

See pricing

Frequently asked questions

Is Hunyuan Hy3 or Hunyuan Hy4 Preview better for coding?

Public SWE-Bench figures are not available for either model, so the honest test is your own repository — run an identical real bug through both. By design, Hunyuan Hy3 leans toward frontier-level reported reasoning and science (gpqa diamond 90.4) at low active-parameter cost while Hunyuan Hy4 Preview leans toward gpqa diamond (92.3), and that positioning usually predicts which feels better on your codebase.

Which is cheaper, Hunyuan Hy3 or Hunyuan Hy4 Preview?

Hunyuan Hy3 is cheaper — Open weight (self-host / free) vs $0.834/$2.501 per 1M tokens.

Which has the bigger context window?

Hunyuan Hy4 Preview — 1M+ tokens vs 256K, about 3.9× larger. Useful only if the model actually reasons over the full window, which not all do.

Should I upgrade from Hunyuan Hy3 to Hunyuan Hy4 Preview?

Since both are Tencent models, the newer one (Hunyuan Hy4 Preview) is usually the better default unless you need a specific cost or latency profile from the other.

Which is newer, Hunyuan Hy3 or Hunyuan Hy4 Preview?

Hunyuan Hy4 Preview — released August 28, 2026, about 53 days after Hunyuan Hy3.

Related comparisons

Specifications and benchmarks reflect publicly reported figures as of June 2026 and may change as providers release updates. Always verify on your own workload.