Hy4
tencent/hy4
Back to model catalog

Tencent · Model guide

Hy4

Compare
tencent/hy4Released Nov 4, 2026

The full release of Tencent's Hy4. Compared with the preview it scales up pre- and post-training, adds image understanding and computer operation, and improves instruction following, long-document comprehension, long-horizon tasks and deep research. It takes text and images over a one-million-token context with up to 64K output tokens, and reasons by default with a direct-answer mode available.

Modalities
TextImageText
Context
1M
Max output
66K
Price / 1M tokens
$ 0.96in$ 2.88out
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.

VendorTencent says Hy4 puts Hy among the strongest open models for the first time, while acknowledging a sizeable gap to the leading closed models.source

Best for
  • Agentic coding · Tencent reports 44.4 on Terminal-Bench 4.0 and 74.9 on DeepSWE 1.1, up from 26.3 and 64.3 for the preview.source

  • Images and computer use · The first Hy model with vision: Tencent highlights fine-grained image work and chart analysis, and trained it to operate software through both the command line and the GUI.(vendor claim)source

  • Finance and office work · Tencent trained on thousands of real finance workspaces and highlights co-editing Word, Excel and PowerPoint files alongside the user.(vendor claim)source

  • Long-context work · The context window is one million tokens, of which up to 960K can be input.source

Neighbouring models
  • tencent/hy3cheaper —Hy3 is the previous generation: a much smaller, text-only model with a 256K context, for high-volume work where Hy4's extra capability is not needed.
Compare them side by side →

Pricing and billing

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

Price / 1M tokensYou pay
Input$ 0.96
Output$ 2.88
Cache read$ 0.048
Cache write$ 0.96
Estimate a request
$ 0.0154
Estimated cost per request
10.0K × $ 0.96
+ 2.0K × $ 2.88
≈ $ 15.36 per 1,000 requests

Capabilities and limits

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

Context window
1.0M tokens
Max output
65.5K tokens
Input / output modalities
Text + Image → Text
Streaming
Supported
Tool calling
Supported
Structured output (JSON)
Supported
Image input
Supported
Thinking mode
Supported· can be turned off
Reasoning effort
none · high
Available endpoints
/v1/chat/completions · /v1/responses · /v1/messages

Published benchmarks

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

Agentic coding
  • Terminal-Bench 4.0
    pass_rate·Tencent Hy · 2026-11-04effort highClaude Codesetup not fully disclosed

    Up to 500 turns and a 13-hour timeout per trial, capped at 16 CPUs and 32 GB RAM; the three GPU-dependent tasks additionally get one H100. The same table quotes Claude Opus 5.5 at 66.4, matching Anthropic's own figure.

    44.4%
Agentic work
  • AutomationBench v1.0.6
    score·Tencent Hy · 2026-11-04effort highsetup not fully disclosed

    The 600 public tasks released by Zapier, scored by the official method.

    56.2%
Coding
  • DeepSWE 1.1
    resolved_rate·Tencent Hy · 2026-11-04effort highmini-swe-agentsetup not fully disclosed

    Each task runs in a sandbox with 8 CPU cores and 16 GB of memory. Tencent reports every model at its highest available reasoning setting, which for Hy4 is high.

    74.9%
  • ProgramBench
    score·Tencent Hy · 2026-11-04effort highClaude Codesetup not fully disclosed

    Almost Solved score, run with Claude Code 2.1.220. Tencent marks ProgramBench as evaluated for the first time at this release; its table quotes DeepSeek-V4.1-Flash-1017 at 29.1, matching DeepSeek's own figure.

    30%
Computer use
  • OSWorld 2.1 (partial credit)
    score·Tencent Hy · 2026-11-04effort highsetup not fully disclosed

    Tencent labels it OSWorld 2.1 (Full, partial score); the same table quotes Claude Opus 5.5 at 81.8 and Kimi K3.1 Pro at 79.6, matching those vendors' own partial-credit figures.

    78.9%
Multimodal reasoning
  • MMMU Pro
    accuracy·Tencent Hy · 2026-11-04effort highsetup not fully disclosed

    Tencent does not say whether tools were available, so this sits under the unqualified MMMU Pro rather than the no-tools or with-tools variants.

    78.6%
  • ZeroBench main (with tools)
    pass_at_5·Tencent Hy · 2026-11-04effort high10 runssetup not fully disclosed

    Network-isolated Jupyter environment that returns image outputs as its tool interface, up to 80 tool-call rounds per run; pass@5 estimated from 10 independent runs.

    55.4%
Reasoning
  • Humanity's Last Exam Full (with tools)
    accuracy·Tencent Hy · 2026-11-04effort highsetup not fully disclosed

    Full set: Tencent flags text-only-subset results with a dagger and this one carries none (the preview's 57.3 in the same row is text-only). The tool set is not disclosed, so it is left unstated.

    59.1%
Scientific research
  • Terminal-Bench-Science 0.1
    accuracy·Tencent Hy · 2026-11-04effort highClaude Codesetup not fully disclosed

    Up to 500 turns and a 13-hour timeout per trial, capped at 24 CPUs and 48 GB RAM.

    19.8%

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.

See all integration guides →

Frequently asked questions

What is Hy4?
Hy4 is available on CrossModel as tencent/hy4. It has a 1M-token context window and can return up to 66K tokens per request.
How much does Hy4 cost?
Currently $ 0.96 per 1M input tokens and $ 2.88 per 1M output tokens. Cached input is billed at $ 0.048 per 1M.
Does Hy4 support tool calling and structured output?
Tool calling: Supported · Structured output (JSON): Supported · Image input: Supported · Streaming: Supported · Thinking mode: Supported
Which endpoint do I call?
For text models, 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.