tencent Open weights Reasoning

Hy3 pricing & benchmarks

Released Jul 6, 2026 · served by 5 providers

Input / 1M tokens$0.129
Output / 1M tokens$0.528
Context window262K
Cached input$0.033
1M in + 300K out$0.29

What Hy3 is

Hy3 is a 295B-parameter Mixture-of-Experts model from Tencent (21B active, 192 experts with top-8 routing) built for reasoning, agentic workflows, and real-world production use. It supports a configurable reasoning effort:...

Providers and prices for Hy3

The same model costs different amounts depending on who serves it. Prices are per 1M tokens, cheapest first. Uptime and throughput are OpenRouter's rolling measurements, not vendor claims.

ProviderInputOutputContext QuantUptime 24hThroughput
Tencentcheapest $0.132 $0.528 262K fp8 99.6%
GMICloud $0.129 $0.534 262K bf16 62.3%
DeepInfra $0.140 $0.580 262K fp8 98.3%
Novita $0.140 $0.580 262K 98.4%
AtlasCloud $0.200 $0.800 262K fp8 99.3%

Hy3 benchmark results

Design Arena head-to-head results, as reported through the OpenRouter model API.

CategoryArenaRankEloWin rate
3d models #39 1227 43.1%
codecategories models #46 1205 41.1%
uicomponent models #49 1201 40.8%
website models #52 1206 41.5%
gamedev models #56 1190 38.6%
dataviz models #68 1155 36.4%

FAQ

How much does the Hy3 API cost?

$0.129 per 1M input tokens and $0.528 per 1M output tokens. A typical workload of 1M input + 300K output tokens costs about $0.29.

Is Hy3 free?

No. It is a paid model starting at $0.129 per 1M input tokens, though some providers offer trial credits.

Which provider is cheapest for Hy3?

Tencent at $0.132 input / $0.528 output per 1M tokens (fp8 quantization).

Can I self-host Hy3?

Yes — weights are published as tencent/Hy3 on Hugging Face, so you can run it on your own hardware.