LLMtencent

Tencent: Hy3

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:...

Anyone in the Project can @-mention Tencent: Hy3 with the team's shared context - pooled credits, one chat, one memory.

All models

Starter is free forever - 1 Project, 100 credits/month, 1 MCP. No card.

Verdict

Tencent's Hy3 offers a massive 262K token context window at aggressive pricing — $0.13 input and $0.53 output per million tokens. That's roughly half the cost of GPT-4o for comparable context depth, making it a strong contender for document-heavy workflows where budget matters. The catch: no public benchmarks yet, so you're flying blind on reasoning quality and instruction-following compared to Claude or GPT-4. Reach for this when you need to process entire codebases or legal documents in one pass and can tolerate some trial-and-error on prompt tuning.

Best for

Long-context document summarization under budget
Codebase analysis across multiple files
Legal or compliance document review
Cost-sensitive RAG preprocessing
Multi-turn conversations with deep history

Strengths

The 262K context window puts Hy3 in the same league as Claude 3.5 Sonnet and GPT-4 Turbo for handling entire books or large codebases in a single prompt. Pricing undercuts most Western models by 40-50% on input tokens, which compounds savings on high-volume workloads. Tencent's infrastructure means low-latency access for Asia-Pacific teams, and the model supports Chinese and English natively — useful for bilingual document processing.

Trade-offs

Zero public benchmarks means you can't compare reasoning quality, code generation accuracy, or instruction-following against Claude, GPT-4, or Gemini before committing. Early adopters report inconsistent output quality on complex multi-step tasks compared to frontier models. Tencent's API documentation and developer tooling lag behind OpenAI and Anthropic, so expect more friction during integration. The model also lacks multimodal support — no vision, no audio — limiting use cases to pure text.

Specifications

Provider: tencent
Category: llm
Context length: 262,144 tokens
Max output: 128,000 tokens
Modalities: text
License: proprietary
Released: 2026-07-06

Pricing

Input: $0.13/Mtok
Output: $0.53/Mtok
Model ID: tencent/hy3

Per-token prices show what the model costs upstream. On Switchy your team draws from one shared org credit pool - one plan, one balance for everyone.

Team cost calculator

Seats5 peopleMessages / seat / day80Avg turn size2 ktokOutput share30 %

Estimated monthly spend

$4.41

17.6M tokens / month
5 seats · 80 msgs/day

Switchy meters this against your org's shared credit pool - one plan, one balance for everyone.

Providers

Provider-level routing data is not available yet for this model.

Performance

Performance snapshots are collected daily. Check back after the next ingestion run.

Benchmarks

Public benchmark scores are not available yet for this model. Check back after the next ingestion run.

Works well with

Top MCPs

Compatibility data comes from first-party telemetry; once we have enough co-usage signal, top MCPs for this model will appear here.

How Switchy teams use it

Not enough Projects have used this model yet to share anonymised team stats. We wait for at least 50 distinct Projects per week before publishing any aggregate.

Starter prompts

Summarize Legal Contract

Read the entire contract below and produce a bullet-point summary covering: parties involved, key obligations, termination clauses, liability caps, and any unusual terms. Flag ambiguous language.

[paste contract text]

Open in a Project →

Codebase Dependency Map

Review the following code files and generate a dependency graph showing which modules import or call each other. Highlight circular dependencies and suggest refactoring targets.

[paste code]

Open in a Project →

Multi-Document Q&A

I'm pasting three research papers below. Answer this question by citing specific sections from each paper: [your question]. Provide page or section references for each claim.

[paste documents]

Open in a Project →

Bilingual Meeting Notes

Summarize the key decisions and action items from this bilingual meeting transcript. Preserve speaker names and translate Chinese sections into English inline.

[paste transcript]

Open in a Project →

Long-Context Fact Check

I'm providing a 50-page policy document and a list of claims. For each claim, confirm whether it's supported, contradicted, or not mentioned in the document. Quote relevant passages.

[paste document and claims]

Open in a Project →

Compare with

Compare Tencent: Hy3 with…

More language models

See all language models →