Skip to content
Model Season every model has a seasonRadarLabs

Hy3 preview

tencent/hy3-preview-20260421
TencentChinaOpen weights

No volume over the last 7 days. Weekly share peaked at 13.7% in the week of May 4, 2026, with 33T tokens processed since Apr 20, 2026.

Launch
Apr 22, 2026
Context
256k tokens
Modality
text → text
Reasoning
yes
License
Open weights
Input
$0.18 / 1M
Output
$0.60 / 1M
Share, last 7 days
—
no volume over the last 7 days
Peak weekly share
13.7%
week of May 4, 2026
Cumulative tokens
33T
since the week of Apr 20, 2026
Weeks in the top 10
10
of 12 weeks with volume

Weekly token share

Share of the router's weekly volume, since the first week with volume

0%5%10%15%20%May '26JunJulAugSepOctpeak 13.7% on May 4, 2026
See the numbers
WeekShare
Oct 19, 20260.000%
Oct 12, 20260.000%
Oct 5, 20260.000%
Sep 28, 20260.000%
Sep 21, 20260.000%
Sep 14, 20260.000%
Sep 7, 20260.000%
Aug 31, 20260.000%
Aug 24, 20260.000%
Aug 17, 20260.000%
Aug 10, 20260.000%
Aug 3, 20260.000%
Jul 27, 20260.000%
Jul 20, 20260.000%
Jul 13, 20260.000%
Jul 6, 20260.777%
Jun 29, 20266.694%
Jun 22, 20267.195%
Jun 15, 20267.774%
Jun 8, 20269.273%
Jun 1, 20268.145%
May 25, 20269.516%
May 18, 202610.617%
May 11, 20269.870%
May 4, 202613.723%
Apr 27, 202612.737%
Apr 20, 20261.539%

Debuted in the week of Apr 20, 2026 and peaked at 13.7% 2 weeks later. No volume was recorded in the latest full week.

What this model is used for new

Tasks where the model ranks among the leaders, and its share of each

This model is not among the leaders in any of the 29 tasks classified in the Oct 25, 2026 snapshot. That does not mean it goes unused: the source lists only the largest models in each task.

Snapshot of the rolling 7-day window through Oct 25, 2026. Share: the model's tokens in the task over the task's total tokens. Weight: the task's share of classified volume.

Where to run it

Endpoints in the source's catalog, from cheapest to most expensive on input

No endpoints listed for this model in the archived catalog.

Benchmarks

Artificial Analysis indexes and benchmarks run by the source, with cost per task

Intelligence
25.3
70th of 199
Coding
58.8
38th of 109
Agents
24.1
35th of 85

The source has not run its own benchmarks on this model.

Artificial Analysis indexes, exposed by the catalog; benchmarks run by the source and archived on Oct 26, 2026. The gray mark is the median of evaluated models.

Against the week's largest models

The five largest proprietary and five largest open-weights models over the last 7 days

ModelLab7-day shareBlended priceContextIntelligenceWeeks in top 10
Proprietary
GPT-6 LunaOpenAIOpenAI4.4%$0.201M38.15
Claude Opus 5.5AnthropicAnthropic4.4%$8.001M57.63
jev-1.13TypesafeTypesafe1.6%———0
GPT-6.1 SolOpenAIOpenAI1.5%$4.001M51.80
Muse Spark 1.3 ContributorMetaMeta1.5%$0.1251M—0
Open weights
DeepSeek V4.1 Flash 1017DeepSeekDeepSeek17.9%$0.2631M39.51
DeepSeek V4.1 FlashDeepSeekDeepSeek9.6%$0.2651M24.77
MiniMax M3.1 FlashMiniMaxMiniMax7.7%$0.2631M42.71
GLM 5.3 FlashZ.ai (GLM)Z.ai (GLM)7.0%$0.2371M41.89
Hy4 previewTencentTencent4.8%$1.251M—8
Hy3 previewthis pageTencentTencent—$0.285256k25.310

Share over the last 7 days (Oct 19, 2026 to Oct 25, 2026). Blended price in US dollars per million tokens, weighted 75% input and 25% output. A dash means the number is not published, not zero.

What this page doesn't say. Share here is a slice of one router's traffic, where the deciding factor is usually price per token. It is not market, revenue or user share, and it does not tell you which model fits your workload.