Tencent Series

Trusted by teams running on

Tencent Cloud

CodeBuddy

Tencent Meeting

WeChat

QQ

Yuanbao

Zhihu

Huawei Cloud

Google Cloud

AWS

Azure

Cloudflare

Alibaba Cloud

The lineup

Choose by output.

Reasoning + vision

Hunyuan 2.0

Fast answers. Deep reasoning.

Tencent's MoE flagship ships in two modes. Instruct for fast answers. Think for deep reasoning. 128K context, same price whether you're drafting or debugging.

Readable typography

HunyuanImage 3.0

Design-grade image output.

Largest open-source image model on the market. Handles readable text, photoreal products, and multi-image fusion. Starts at $0.011 per image.

Cinematic motion

HunyuanVideo 1.5

Prompt-to-video motion.

8.3B DiT that runs on a consumer GPU. Text-to-video, image-to-video, native 720p, upscalable to 1080p. HN called it 'the Flux moment for video.'

Prompt to GLB

Hunyuan3D V3.0

Prompt/photo to 3D.

Text, image, or sketch-to-3D through Tencent HY 3D Global. Generates textured geometry, PBR materials, GLB/OBJ exports, and smart topology for engine-ready meshes.

Agentic coding

Kimi K2.6

Agentic coding at cost.

1T-param MoE from Moonshot, a Tencent-backed lab. Tops SWE-Bench Pro (58.6 — beats Claude Opus 4.6 and Gemini 3.1 Pro). 262K context, 300-agent swarms, 12-hour autonomous sessions.

Teams

Ship more. Spend less.

Shoot a product once. Generate 50 store photos, lifestyle shots, and ad variants in an afternoon.

“Replaced our studio calendar. Same-day assets instead of two-week lead times, 80% cheaper per SKU.”

Mei Tanaka, Head of Store Creative

Localize a campaign across 15 markets — rewrite copy, swap visuals, re-cut video — in one afternoon.

“Shipped 15 localized launch assets before dinner. Used to take us a week and three agencies.”

Rafael Cruz, Performance Marketing Lead

Point Kimi K2.6 at a legacy repo. Spin up a 30-agent swarm to split bug fixes, refactors, and test generation in parallel.

“K2.6 landed a three-day refactor in one overnight run. I reviewed the PR over coffee.”

Liang Wu, Principal Engineer

Photograph a stack of invoices. Get vendor, line items, totals, tax, and due dates as structured JSON.

“Monthly close moved up three days. 500 invoices in twenty minutes instead of two people for two days.”

Sven Eriksson, Finance Operations

Turn a selfie into a 3-second animated Reel intro. Batch ten thumbnail concepts before posting.

“Every Reel gets a cinematic intro now. Watch-time up 25% in the first month.”

Ayaka Mori, Creator

Turn a brief into 40 visual concepts in an hour. Apply the brand's illustration style to 50 new assets via API.

“Brought 40 concepts to the first client meeting. They signed the retainer that day.”

Anika Shah, Brand Designer

Customer sends a support screenshot. The model reads the error, identifies the issue, and drafts a reply with exact steps.

“First-response dropped from eight minutes to ninety seconds. We handle triple the ticket volume per shift.”

Zara Ahmed, Support Lead

Turn concept sketches and product references into GLB assets, then clean topology for game-engine handoff.

“We go from moodboard to textured prop in the same review cycle. The mesh is clean enough for the engine team to keep moving.”

Thandi Nkosi, 3D Production Lead

Photograph an empty room. Stage it in three styles — modern, mid-century, family — at cents per image.

“Virtual staging at $0.05 a room instead of $3,000. Listings moved 40% faster.”

Hassan Karim, Listing Agent

Describe your business once. Get 20 logo concepts, a 30-day content calendar, and four short video ads.

“Entire brand plus a month of content in one Sunday. Under ten dollars on BasicRouter.”

Joon-ho Park, Founder

Use cases

Prompt. Build. Ship.

Describe the scene. HunyuanVideo ships it.

Write a prompt — product demo, brand story, concept test — and get a 5-second cinematic clip with camera movement, lighting, and motion. Native 720p, upscalable to 1080p.

hunyuan-video-1.5

Why BasicRouter

Tencent rates, without Tencent setup.

One API for every Hunyuan model + Kimi K2.6

Tencent Cloud infrastructure partner

What you get

Tencent backbone
OpenAI drop-in
One key, one bill
Native rates

Tencent Cloud Partner

Trusted by 150+ enterprises on the backbone behind WeChat & WeChat Pay.

Native pricing, no markup

Up to 10× cheaper than other resellers — match Tencent's official rates exactly.

Pricing

Native rates, no reseller markup.

Hunyuan Text · 5 models

/ 1M input tokens · from

✓hunyuan-turbos — $0.11 in / $0.28 out per 1M
✓hunyuan-t1 — $0.14 in / $0.55 out per 1M
✓hunyuan-2.0-instruct — $0.44 in / $1.10 out per 1M
✓hunyuan-2.0-thinking — $0.55 in / $2.19 out per 1M
✓hunyuan-turbos-vision — $0.42 in / $1.24 out per 1M

HunyuanImage 3.0 · 3 variants

/ image (1MP) · from

✓hunyuan-image-3.0 — $0.011 / image (1MP)
✓hunyuan-image-3.0-instruct — $0.015 / edit
✓hunyuan-image-3.0-instruct-distil — $0.008 / image (8-step fast)
✓Up to 4K output, prompt + image-conditioned

HunyuanVideo 1.5 · t2v + i2v

/ second @ 720p · official Tencent rate

✓hunyuan-video-1.5 t2v — $0.015 / second
✓hunyuan-video-1.5 i2v — $0.015 / second
✓720p and 1080p, 5–10 second clips
✓≈ 5× cheaper than reseller per-second listings

Hunyuan3D V3.0 · Express + Pro

/ Express generation · from

✓hunyuan3d-v3.0 express — 15 credits / ≈ $0.23
✓hunyuan3d-v3.0 professional — 25 credits / ≈ $0.38
✓PBR / multiview / custom face count — +10 credits
✓smart topology post-process — 50 credits

Kimi K2.6 · coding agent

/ 1M input tokens · native

✓kimi-k2.6 — $0.95 in / $4.00 out per 1M
✓cache hit — $0.16 / 1M (≈ 6× cheaper)
✓OpenAI-compatible chat + tool-use endpoint
✓Same price whether you swarm 1 or 300 sub-agents

What $5 gets you before you commit budget.

$5 covers roughly 450 HunyuanImage generations, 66 seconds of HunyuanVideo, 13 professional Hunyuan3D textured assets, or a full overnight Kimi K2.6 agent run.

450 poster variations

Batch 450 HunyuanImage 3.0 generations at $0.011 each. Pick the winners, discard the rest — all under one free credit.

A minute of finished video

Turn 12 product stills into 5-second HunyuanVideo 1.5 clips. Ship to TikTok, Reels, and store pages before the meeting ends.

13 textured 3D assets

Generate a small prop set with Hunyuan3D V3.0 Professional. Export GLB/OBJ previews, then spend extra credits only on the winners.

One overnight agent swarm

Fire off a 300-sub-agent Kimi K2.6 run on a legacy refactor. Wake up to a ready PR — for less than the cost of a coffee.

500 pages analyzed

Drop a 500-page report into HY 2.0 Think. Get structured findings, contradictions flagged, and an executive summary back.

A full campaign preview

Generate four concepts, convert each to a 5-second teaser, write three ad copy variants per concept — all before committing real spend.

FAQ

What teams ask before launch.

What does $5 of credit get me?

Roughly 450 HunyuanImage 3.0 generations, 66 seconds of HunyuanVideo 1.5, 13 professional Hunyuan3D V3.0 textured assets, a full overnight Kimi K2.6 agent run, or thousands of Hunyuan-TurboS text calls. Enough to stress-test every model end-to-end.

How do your prices compare to other resellers?

We're a Tencent Cloud infrastructure partner with direct access to model serving. On most models we match Tencent's official rates. On select models — HunyuanImage 3.0 especially — we're up to 10× cheaper than resellers who mark up the same weights on Fal.ai, Replicate, or AIMLAPI.

Which model should I try first?

hunyuan-turbos for general-purpose chat at $0.11/1M input — ridiculously cheap. hunyuan-image-3.0 if you make visuals. kimi-k2.6 if you code. Start wherever your use case lives.

When do I use HY 2.0 Instruct vs Think?

Instruct for fast answers, drafts, customer-facing chat, summarization. Think for reasoning, planning, multi-step analysis, or anywhere the answer has to be right. Same API, switch with one flag.

Is Kimi K2.6 really from Tencent?

Kimi is built by Moonshot AI. Tencent co-leads Moonshot's funding and ships Kimi as the coding backend inside Tencent CodeBuddy. We deliver it at Moonshot's native price — no aggregator markup.

What does Hunyuan3D V3.0 export?

Text, image, or sketch input can become textured geometry with PBR material options. Export GLB/OBJ for review, then use smart topology or UV tools when the asset needs to move into Unity, Unreal, Blender, or 3D printing.

Why not call Tencent Cloud directly?

You can. But then you manage five separate model endpoints, five billing meters, and five sets of credentials. BasicRouter is one OpenAI-compatible endpoint, one key, one bill — and we match Tencent's official rates, so the API convenience is free.

Is this production-ready?

150+ enterprises run production workloads on our infrastructure today. Built on Tencent Cloud, with AWS, Azure, Google Cloud, and Cloudflare at the edge. Same SLA as calling Tencent directly.

Start building today.

$5 free credit — enough for 450 images, 66 seconds of video, 13 professional Hunyuan3D assets, or a full overnight coding agent run. No card required.

All of Tencent AI.Under onekey