Tasks and inputs
Hy4 Flash is listed for Image-text-to-text, Text generation. It accepts text, image and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
Tencent Hy4 Flash is the cost-focused companion to Hy4, keeping vision and agent skills at a far lower token price. Availability is still unsettled.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Pricing conditions: Indicative USD per 1M tokens. OpenRouter (Tencent Cloud): $0.035 input / $0.1405 output, likely a 50% launch discount; reseller list about $0.07-0.08 / $0.28-0.32. No Tencent price page read. Confirm on your own account.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
Cheap is the point. When Tencent first named this model at its Global Digital Ecosystem Conference on 29 October 2026, the pitch was simple: keep Hy4's visual understanding and agent capabilities, cut the cost. Everything else about it was, at the time of writing, still settling.
We are publishing this with low confidence, and you deserve the reasons up front.
tencent/hy4-flash, served by Tencent Cloud, with live per-token prices.So: it is probably live, possibly in a staged rollout. One test call with your own key will tell you more than any of these sources.
| Source | Input (USD per 1M) | Output (USD per 1M) |
|---|---|---|
| OpenRouter, Tencent Cloud provider | $0.035 | $0.1405 |
| Reseller list prices | about $0.07–0.08 | about $0.28–0.32 |
The OpenRouter figure is exactly half the reseller range, which matches the 50% launch discount Tencent applied to Hy4 on TokenHub. That is our inference, not a published statement. We did not read a Tencent price page for Flash. Treat every number as indicative and check your account.
Even at list, that is roughly a tenth of Hy4's per-token price; both models are on the Tencent provider page.
Text and image input, text output. Model listings give a 1,048,576-token context and a 65,536-token output ceiling, matching Hy4. OpenRouter's interface accepts tools, tool_choice, response_format and reasoning parameters. Tencent has not published a parameter count or benchmark scores for Flash that we could confirm.
Think high-volume, lower-stakes work. Screenshot triage. Document classification. First-pass extraction before a stronger model checks the hard cases. Routing layers in an agent stack.
The older Hy3 is another baseline if you already run it.
Compare the model on the work you need to complete.
Hy4 Flash is listed for Image-text-to-text, Text generation. It accepts text, image and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
Documented limits: 1.05M context and 65.54K maximum output. The listed route has starting rates of $0.07 input and $0.281 output per million tokens. Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.
Use the model identifier and account requirements for the selected API route. Confirm its request limits and output format with one complete task before connecting a larger workflow.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
Alibaba
Qwen4 Max (qwen4-max) is Alibaba's first Qwen4-generation flagship: 1M-token context, text, image and video input, thinking on by default, $1.50/$4.50 per 1M.
Sarvam
Sarvam-105B-1711 is a refreshed open post-train of Sarvam's 105B MoE, tuned for tool calling, instruction following, agents and Indian languages.
IBM
Granite 4.1 8B LLMON Preview is IBM's model trained on block-structured LLMON prompts to resist indirect prompt injection in agents and RAG.
NVIDIA
Nemotron-Labs Sigma 8B is NVIDIA's open continuous diffusion language model, denoising 64-token blocks in parallel; English, GPU-only, 8K context.
Alibaba
Qwen4 Max Preview was Alibaba's five-day early-access build of its Qwen4 flagship, offered only through a Token Plan subscription before GA qwen4-max.
speakleash
Bielik-11B-v3.1-Instruct is SpeakLeash's open Polish-first 11B instruct model, trained on ~56 languages, with five reasoning levels and tool calling.