Hy3 Preview on VM0. Tencent's low-cost reasoning preview
Tencent's text-only OpenRouter preview model on VM0. 256K context, configurable reasoning, and cost-saving x0.1 VM0 Managed pricing.
256K tokens · Text / Code · Prompt cache
Hy3 Preview is Tencent's efficient text model exposed on VM0 through OpenRouter. It offers a 256K-token context window and a very low provider price, making it useful for high-volume text and code workflows where Sonnet-level reasoning is not required.
Use it for budget-sensitive routing, summaries, extraction, and first-pass code review. Keep in mind that it is a preview model and text-only, so it should not be the default for vision tasks or high-stakes reasoning.
What is Hy3 Preview?
June 2026 · Tencent preview model exposed through VM0 Managed via OpenRouter.
Hy3 Preview is a Tencent model available on OpenRouter. VM0 exposes it as the canonical hy3-preview model and routes Built-in runs to tencent/hy3-preview.
The model focuses on efficient text reasoning with a 256K-token context window. It supports OpenRouter reasoning controls and prompt cache reads.
On VM0 Managed it sits at x0.1 credits. Teams can also bring an OpenRouter API key and select the upstream id directly.
What's notable about Hy3 Preview
Headline architecture and capability features.
Hy3 Preview is exposed on VM0 through the OpenRouter Anthropic-compatible gateway with upstream id tencent/hy3-preview.
Specs at a glance
Hy3 Preview benchmarks
Hy3 Preview is positioned as a low-cost text route. Because preview-model behavior can change, VM0 keeps this page focused on operational fit rather than fixed leaderboard claims.
Hy3 Preview pricing
Provider list price, per 1M tokens.
How Hy3 Preview behaves in practice
Observed behaviour from production agent runs.
Text throughput
Hy3 Preview is best used for repeated text-heavy steps where cost control matters more than premium reasoning depth.
Preview behavior
Treat it as a model to A/B on real workflows. Keep critical long-running agents on more proven routes until the task profile is validated.
Routing
VM0 Managed routes through OpenRouter using tencent/hy3-preview. OpenRouter BYOK users can select the same upstream model directly.
Best agent tasks for Hy3 Preview
High-volume summarization
Use Hy3 Preview for repeated summaries, ticket digests, and internal-note condensation where x0.1 credits keeps volume economical.
Extraction and classification
Run structured extraction or triage over batches of text before escalating ambiguous cases to Sonnet or Opus.
Budget code review pass
Use it for first-pass code review or diff explanation when the goal is broad coverage and the findings will be checked downstream.
When to skip Hy3 Preview
Skip Hy3 Preview for image, audio, or video input, and for workflows where preview-model variance is not acceptable.
Hy3 Preview vs other models
Hy3 Preview vs MiMo-V2.5
MiMo-V2.5 has 1M context and multimodal inputs. Hy3 Preview is narrower, text-only, and aimed at efficient text reasoning.
Hy3 Preview vs GLM-5.2
GLM-5.2 has the larger 1M window and a Z.AI route. Hy3 Preview is the lower-cost Tencent/OpenRouter option for standard long-context text.
Hy3 Preview vs DeepSeek V4 Pro
Both are cost-saving text/code routes. DeepSeek V4 Pro is the more established cost-first model in VM0; Hy3 Preview is useful for Tencent/OpenRouter A/B tests.
Hy3 Preview vs Claude Sonnet 4.6
Sonnet is the stronger default for high-confidence tool use and reasoning. Hy3 Preview is a budget route for less risky text work.
Bottom line: should you use Hy3 Preview?
Pick Hy3 Preview for low-cost OpenRouter text workloads that can tolerate preview-model behavior.
Frequently asked questions
Is Hy3 Preview available through VM0 Managed?
Yes. VM0 Managed routes Hy3 Preview through OpenRouter with the upstream id tencent/hy3-preview.
Does Hy3 Preview support image input?
No. OpenRouter lists Hy3 Preview as text input and text output.
Can I use my own OpenRouter key?
Yes. Select the OpenRouter provider and use the upstream model id tencent/hy3-preview.
Alternatives
Using Hy3 Preview on VM0
Two ways to access Hy3 Preview on VM0
VM0 supports Hy3 Preview as a Built-in model billed in VM0 credits, and through bring-your-own with a OpenRouter API key. The Built-in path uses VM0 Managed routing and the credit multiplier explained below; the bring-your-own path bills you directly with the upstream vendor and skips the VM0 credit conversion entirely.
VM0's recommendation
VM0 positions Hy3 Preview as a cost-saving option rather than a core agent model. Use it to optimise unit cost on non-core work, such as bulk classification, pre-filters, latency-critical short replies, or pinned legacy agents, while keeping Claude Opus 4.7, Claude Opus 4.6, or Claude Sonnet 4.6 on the steps that decide the run.
Credits and the ×0.1 multiplier
Every Built-in model on VM0 is priced as a multiple of Claude Sonnet 4.6, which sits at the ×1 credit baseline. Hy3 Preview bills at ×0.1 credits. The multiplier is what shows up on your VM0 invoice; the vendor list price in the pricing table above is what the upstream provider charges before VM0 converts it into credits.
Hy3 Preview bills at ×0.1, which means a step here costs only 0.1× the credits of an equivalent step on Sonnet 4.6 (the ×1 baseline). That puts it well below the credit baseline and makes it the natural pick for high-volume background work where cost-per-step matters more than peak reasoning quality.
Available on VM0 since June 2026.