CrossModel
Hy3
tencent/hy3
Back to model catalog

Tencent · Model guide

Hy3

Compare
tencent/hy3

Tencent's open 295B-parameter MoE for reasoning and agent workflows, activating 21B parameters per token. It supports a 256K context window, selectable direct or deep reasoning, and production-oriented tool use.

Modalities
TextText
Context
262K
Max output
262K
Price / 1M tokens
$ 0.16in$ 0.64out
StreamingTool callingJSON outputImage inputThinking mode · can be turned off

What this model is good at

What the vendor positions it for, and which of our models to reach for instead.

VendorA cost-efficient reasoning and agent model that rivals much larger open flagships.source

Best for

  • Agentic coding · Official evaluations cover repository repair, terminal operation, MCP tools and production coding tasks.source

  • Productivity agents · Tencent highlights coding, office work, financial modeling and frontend design.(vendor claim)source

Pricing and billing

Billed per token. Cached input is charged at the cache-read rate.

Price / 1M tokensYou pay
Input$ 0.16
Output$ 0.64
Cache read$ 0.040
Cache write$ 0.16
Estimate a request
$ 0.0029
Estimated cost per request
10.0K × $ 0.16
+ 2.0K × $ 0.64
≈ $ 2.88 per 1,000 requests

Capabilities and limits

What CrossModel guarantees across every route this model can take right now.

Context window
262.1K tokens
Max output
262.1K tokens
Input / output modalities
Text → Text
Streaming
Supported
Tool calling
Supported
Structured output (JSON)
Supported
Image input
Not supported
Thinking mode
Supported· can be turned off
Reasoning effort
none · low · high
Available endpoints
/v1/chat/completions · /v1/responses · /v1/messages
First-party sources:Specsgithub.com 1tencent.com 2Capabilitiesgithub.com

Published benchmarks

Scores the vendor reported, with the evaluation setup each one came from.

Agentic coding
  • Terminal-Bench 2.1
    pass_rate·Tencent Hy · 2026-07-06effort highTerminus-2setup not fully disclosed
    71.7%
Agentic search
  • BrowseComp (with tools)
    accuracy·Tencent Hy · 2026-07-06effort highsetup not fully disclosed

    Tencent used an internal harness with self-summary context management.

    84.2%
Agentic tasks
  • Claw-Eval General
    pass3_rate·Tencent Hy · 2026-07-06effort highsetup not fully disclosed

    105-query internal harness evaluation; pass^3.

    68.5%
Coding
  • SWE-Bench Multilingual
    resolved_rate·Tencent Hy · 2026-07-06effort highSWE-agentsetup not fully disclosed
    75.8%
  • SWE-Bench Pro
    resolved_rate·Tencent Hy · 2026-07-06effort highSWE-agentsetup not fully disclosed
    57.9%
  • SWE-Bench Verified
    resolved_rate·Tencent Hy · 2026-07-06effort highsetup not fully disclosed

    Tencent reports less than four percentage points of variance across CodeBuddy, Cline and KiloCode scaffoldings.

    78%
Long context
  • AA-LCR
    score·Tencent Hy · 2026-07-06effort highsetup not fully disclosed
    73.4%
Reasoning
  • Humanity's Last Exam (with tools)
    accuracy·Tencent Hy · 2026-07-06effort highsetup not fully disclosed

    Text-only subset.

    53.2%
Scientific reasoning
  • GPQA Diamond
    accuracy·Tencent Hy · 2026-07-06effort highsetup not fully disclosed
    90.4%
Tool use
  • MCP Atlas Public
    score·Tencent Hy · 2026-07-06effort highsetup not fully disclosed

    Official April 2026 methodology with a 500-task public subset.

    79.1%

Vendor-reported numbers, not CrossModel measurements. Scores are only comparable when the benchmark version, metric and evaluation setup match, so nothing here is averaged or ranked.

Use it in your tools

Point the base URL at CrossModel and paste this model ID — every tool below has a setup guide.

Claude Code
Anthropic's official command-line coding assistant
Cursor
AI-first code editor
Cline
Open-source AI coding agent for VS Code
Codex
OpenAI's coding agent for ChatGPT desktop and the terminal
OpenCode
Open-source terminal coding assistant
Zed
High-performance, AI-native code editor
Open WebUI
Open-source, self-hosted AI chat interface
Continue
Open-source AI coding assistant for VS Code & JetBrains
See all integration guides →

Frequently asked questions

What is Hy3?
Hy3 is available on CrossModel as tencent/hy3. It has a 262K-token context window and can return up to 262K tokens per request.
How much does Hy3 cost?
Currently $ 0.16 per 1M input tokens and $ 0.64 per 1M output tokens. Cached input is billed at $ 0.040 per 1M.
Does Hy3 support tool calling and structured output?
Tool calling: Supported · Structured output (JSON): Supported · Image input: Not supported · Streaming: Supported · Thinking mode: Supported
Which endpoint do I call?
OpenAI Chat Completions (/v1/chat/completions), OpenAI Responses (/v1/responses) and Anthropic Messages (/v1/messages) all work — pick whichever your SDK already speaks.
Can I try it without writing code?
Yes. Sign in and open the Playground on this page to chat with the model directly — no API key needed. Usage is billed from your wallet at the normal rate.

Chat first, integrate later

Chat first, then wire it in once you like the answers.