Volcengine API
Access Volcengine instantly with Puter.js, and add AI to any app in a few lines of code without backend or API keys.
// npm install @heyputer/puter.js
import { puter } from '@heyputer/puter.js';
puter.ai.chat("Explain AI like I'm five!", {
model: "volcengine/doubao-seed-2.0-pro"
}).then(response => {
console.log(response);
});
<html>
<body>
<script src="https://js.puter.com/v2/"></script>
<script>
puter.ai.chat("Explain AI like I'm five!", {
model: "volcengine/doubao-seed-2.0-pro"
}).then(response => {
console.log(response);
});
</script>
</body>
</html>
List of Volcengine Models
Doubao Seed 1.8
volcengine/doubao-seed-1.8
Doubao Seed 1.8 is a generalized agentic model from ByteDance's Seed team, released in December 2025 as the predecessor to the Seed 2.0 family. It combines native vision with text input and offers three adjustable thinking modes so callers can trade off latency against reasoning depth. The model targets search and agent workflows such as information retrieval, GUI interaction, financial analysis, and customer service. On BrowseComp-en it scores 67.6, ahead of Gemini 3 Pro, and it reaches 87.8 on VideoMME for video understanding. ByteDance also reports it matches Seed 1.6's intelligence using roughly a third of the tokens, 5K versus 15K, improving token efficiency for search-heavy tasks. Independent evaluations note it trails Gemini 3 and GPT-5 on general benchmarks, making it a better fit for search and agentic use cases than coding. Its 32,000-token output limit is smaller than the 128,000 available on the Seed 2.0 tiers.
ChatDoubao Seed 2.0 Code
volcengine/doubao-seed-2.0-code
Doubao Seed 2.0 Code is ByteDance's coding-specialized model in the Seed 2.0 family, released in February 2026 alongside Pro, Lite, and Mini. It's built for agentic software development rather than single-turn code completion, breaking a task into a plan and revising that plan as it executes. The model reads screenshots, diagrams, and charts alongside code, so design-to-code and chart-driven development requests can include visual context. It supports a 256,000-token context window and integrates with ByteDance's TRAE IDE for in-editor agentic coding. ByteDance has not published standard coding benchmark scores, such as SWE-bench or LiveCodeBench, specifically for this variant, so third-party numbers circulating for it should be treated cautiously. It's built for multi-file bug fixes, refactors, and IDE-integrated agentic coding workflows rather than general chat or reasoning tasks.
ChatDoubao Seed 2.0 Lite
volcengine/doubao-seed-2.0-lite
Doubao Seed 2.0 Lite is ByteDance's balanced, production-grade tier in the Seed 2.0 family, positioned between the flagship Pro model and the lighter Mini tier. ByteDance describes it as trading off output quality and response speed for high-frequency enterprise use rather than maximum reasoning depth. It handles unstructured information processing, content creation, search and recommendation, and data analysis. An April 2026 update added omni-modal understanding across video, image, audio, and text, plus GUI capabilities that let it identify buttons, menus, and forms and carry out browser or computer actions. On benchmarks it scores 93.0 on AIME 2025, a 2233 Codeforces rating, 81.7 on LiveCodeBench v6, and 73.5 on SWE-Bench Verified, and it edges out the larger Pro model on MMLU-Pro (87.7 versus 87.0). ByteDance reports the updated Lite model outperforms GPT-4o mini and Gemini 3 Flash on numerous evaluations.
ChatDoubao Seed 2.0 Mini
volcengine/doubao-seed-2.0-mini
Doubao Seed 2.0 Mini is the smallest and cheapest model in ByteDance's Seed 2.0 family, built for low-latency, high-concurrency, cost-sensitive deployments rather than peak reasoning quality. It still supports multimodal input (text, image, and video) and the adjustable reasoning modes shared across the Seed 2.0 line. Benchmark scores trail the larger tiers, landing at 87.0 on AIME 2025, a 1644 Codeforces rating, 64.1 on LiveCodeBench v6, 67.9 on SWE-Bench Verified, and 81.2 on VideoMME. At roughly $0.04 per million input tokens and $0.32 per million output tokens, Mini is the least expensive of the five Seed models here, priced for batch classification, content moderation, high-volume customer service, and other workloads where throughput and cost matter more than top-tier reasoning.
ChatDoubao Seed 2.0 Pro
volcengine/doubao-seed-2.0-pro
Doubao Seed 2.0 Pro is ByteDance's flagship reasoning model in the Seed 2.0 family, built for long-chain reasoning and complex, multi-step workflows. It scores 98.3 on AIME 2025, 88.9 on GPQA Diamond, and 97.3 on HMMT, and ByteDance reports gold medal-level results on the IMO, CMO, and ICPC competitions. On coding it reaches a 3020 Codeforces rating, 87.8 on LiveCodeBench v6, and 76.5 on SWE-Bench Verified. On agentic and multimodal tasks it posts 77.3 on BrowseComp, 55.8 on Terminal-Bench, and 89.5 on VideoMME for hour-long video analysis. Independent reviews put its AIME 2025 score above GPT-5.2's reported 93 and Gemini 3 Pro's 87, and price it well below GPT-5.2 and Claude Opus 4.5 on both input and output tokens. It's the tier to reach for math, competitive programming, agent, and video-understanding tasks where Lite and Mini fall short.
Frequently Asked Questions
The Volcengine API gives you access to models for AI chat. Through Puter.js, you can start using Volcengine models instantly with zero setup or configuration.
Puter.js supports a variety of Volcengine models, including Doubao Seed 1.8, Doubao Seed 2.0 Code, Doubao Seed 2.0 Lite, and more. Find all AI models supported by Puter.js in the AI model list.
With the User-Pays model, users cover their own AI costs through their Puter account. This means you can build apps without worrying about infrastructure expenses.
Puter.js is a JavaScript library that provides access to AI, storage, and other cloud services directly from a single API. It handles authentication, infrastructure, and scaling so you can focus on building your app.
Yes — the Volcengine API through Puter.js works with any JavaScript framework, Node.js, or plain HTML. Just include the library and start building. See the documentation for more details.