Questions? hello@viberation.devGet supportBlogDocsChangelog
Get started

Tencent's open-weight models, aimed at coding agents and tool use.

Hy is Tencent's open-weight family, successor to its Hunyuan models. The current preview is a mixture-of-experts model built for coding agents and complex tool-use workflows with a million-token context, and the line also carries small translation-specialised models you can run almost anywhere. Preview status is worth taking literally: good for trying, not for depending on.

At a glance — Hy3 preview

Cost
$0.18 in · $0.60 out per 1M tokensBudget
Memory
262K tokens — about 400 pages
Reads
Text
Can do
  • Use tools (agents)
  • Structured output (not supported)
  • Thinking mode
Good at
Coding: better than 67% · Agentic work: better than 60% · Overall: better than 51%
Reliability
99.96% uptime, last 24h
Released
Apr 22, 2026
Advanced details
API model ID
tencent/hy3-preview
Max output
236K tokens
Cached input
$0.06 per 1M
Tokenizer
Other
Coding index
58.8
Agentic work index
24.1
Overall index
25.3

Providers

The same model, hosted by different companies. OpenRouter picks one per request and falls back to the next if it fails.

ProviderInput /1MOutput /1MCached /1MMax outputUptime 24h
GMICloud · bf16$0.18$0.60$0.06236K99.96%

Supported parameters

  • include_reasoning
  • max_tokens
  • reasoning
  • reasoning_effort
  • seed
  • temperature
  • tool_choice
  • tools
  • top_p

Live from OpenRouter, refreshed hourly. Scores from Artificial Analysis.

Key info

Pricing
Open source
Category
Models
Best for
Expert
Made by
Tencent
Weights
Open
Good for
Coding agents and tool-use workflows

More in Models

  • France's frontier lab, with the small models everyone self-hosts.

    Models
  • Meta's current line, built for long multi-agent runs.

    Models
  • The open-weight family everyone else gets benchmarked against.

    Models
  • Google's open models — Gemini's research, small enough to self-host.

    Models