← All models

Seed2.0 Mini

Published Publisher ByteDancedoubao-seed-2-0-mini-260215

Overview

Seed2.0 Mini is the smaller general-purpose agent model in the Seed2.0 series, intended for low-latency, high-concurrency and cost-sensitive workloads. The official series announcement is dated February 14, 2026; it is a series announcement, not independent confirmation of the exact release date of this API snapshot. The 260215 suffix is an identifier and is not used here to infer a release date. (seed.bytedance.com)

Capabilities and uses

The corresponding BytePlus seed-2-0-mini-260215 listing describes deep reasoning, text generation, multimodal understanding, tool use and structured output. The domestic documentation also lists the specified version for image understanding. These capabilities support applications such as assistants, document or image analysis, and agent workflows. (docs.byteplus.com)

Limits and availability

  • The BytePlus listing specifies a 256K-token context, a maximum input of 224K tokens and a maximum output of 128K tokens, including reasoning tokens. Platform and API limits should be checked before use. (docs.byteplus.com)
  • The weight availability status of this specific version is unverified. Availability through a hosted API alone does not establish whether its weights are open or closed.
  • Volcano Engine scheduled doubao-seed-2-0-mini-260215 for service termination on October 22, 2026 at 14:00 UTC+8, with doubao-seed-2-0-mini-260428 recommended for migration. BytePlus separately scheduled its seed-2-0-mini-260215 version for retirement on November 11, 2026. Check the relevant platform before relying on an existing endpoint; a replacement can differ in behavior and parameters. (docs.volcengine.com)

Official pricing

Checked November 17, 2026. These are separately published platform rates, not currency conversions or necessarily interchangeable offerings. Since the listed snapshots have reached their announced retirement dates, the rates are reference prices, not a guarantee of current availability.

  • China — Volcano Engine: Its official Lake AI Service billing documentation lists the exact domestic snapshot under Ark standard online, pay-as-you-go inference. For non-audio input, prices in CNY per million tokens are: input length 0–32K, input ¥0.20 and output ¥2.00; over 32K–128K, input ¥0.40 and output ¥4.00; over 128K–256K, input ¥0.80 and output ¥8.00. The input-length tier determines the rate. Official domestic pricing. (docs.volcengine.com)
  • International — BytePlus ModelArk: The official retiring-model table lists the international ID seed-2-0-mini-260215, not the domestic doubao- ID. For standard inference, prices in USD per million tokens are: prompt length 0–128K, input $0.10 and output $0.40; over 128K–256K, input $0.20 and output $0.80. These are the listed standard rates, not batch or cached-input rates. Official international pricing. (docs.byteplus.com)

Official sources