How much does the ByteDance Seed API cost?

Seed 2.1 Turbo costs $0.50 per 1M input tokens, $0.10 cached and $2.50 output on BytePlus ModelArk. The lineup runs from $0.10 to $0.50 input, as of 2026-10-30.

What is the cheapest Seed model?

Seed 2.0 Mini at $0.10 input and $0.40 output per 1M tokens, or half that on Flex.

Does Seed 2.0 Pro cost more for long prompts?

Yes. Above 128K tokens it costs $1 / $6 per 1M tokens, double the standard price. Seed 2.1 Turbo keeps one price up to 256K.

Which Seed models are being retired?

Seed 1.6, Seed 1.6 Flash, Seed 1.8 and the older dated builds of Seed 2.0 Lite and Mini retire on Nov 11, 2026, along with two third-party models on BytePlus. BytePlus lists a replacement for each.

What is BytePlus Flex pricing?

A cheaper tier at about half the standard input and output price, with cached input at the standard rate. Seed 2.1 Turbo costs $0.25 in and $1.25 out on Flex.

Is Seed the same as Doubao?

Yes, it's the same ByteDance model family. Doubao is the name on Volcano Engine in mainland China (priced in yuan); Seed is the name on BytePlus internationally (priced in dollars).

ByteDance Seed API pricing: Seed 2.1 Turbo, 2.0 Pro, Flex

USD per 1M tokens, BytePlus ModelArk standard (online) list price, checked 2026-10-30. Flex is BytePlus' discounted tier; Seed 2.0 Code Preview has no Flex price.

Model	Input	Cached	Output	Above 128K (in / out)	Flex (in / out)	Context
Seed 2.1 Turbo	$0.50	$0.10	$2.50	same price	$0.25 / $1.25	256K
Seed 2.0 Pro	$0.50	$0.10	$3	$1 / $6	$0.25 / $1.50	256K
Seed 2.0 Code Preview	$0.50	$0.10	$3	$1 / $6	n/a	256K
Seed 2.0 Lite	$0.25	$0.05	$2	$0.50 / $4	$0.125 / $1	256K
Seed 2.0 Mini	$0.10	$0.02	$0.40	$0.20 / $0.80	$0.05 / $0.20	256K

Which ByteDance Seed model should you use?

Seed 2.1 Turbo for agents and general work. Seed 2.1 Turbo is the newest model, one price across its 256K window, and the only Seed model that can write up to 256K tokens in one answer, per BytePlus' model list. LMArena doesn't rank it yet.
Seed 2.0 Pro if you want a ranked model. Seed 2.0 Pro is #28 of 51 priced models on Text Arena, at $0.50 in and $3 out. Seed 2.1 Turbo is cheaper on output ($2.50) and newer, so Pro mostly makes sense if you already run it.
Seed 2.0 Code Preview for coding experiments. Seed 2.0 Code Preview matches Pro's price, but it's a preview and BytePlus lists no MCP support and no structured output for it.
Seed 2.0 Lite and Mini for bulk work. Seed 2.0 Lite is $0.25 / $2 and Seed 2.0 Mini is $0.10 / $0.40, number 4 of 77 on our full price list.

How Flex and Batch pricing work on BytePlus

BytePlus sells the same models three ways. Standard is the price in the table. Flex charges about 50% of standard for input and output, per BytePlus' billing rules, while cached input stays at the standard price; Seed 2.1 Turbo works out to $0.25 in and $1.25 out. Flex only supports implicit caching and has no cache-storage fee.

Batch inference is listed only for Seed 2.0 Lite and Mini among the Seed models, at the same half price as Flex. Standard requests that use explicit caching also pay a cache-storage fee of $0.0083 per 1M tokens per hour.

The 128K price step

Every Seed 2.0 model doubles in price once a prompt passes 128K tokens. A 200K-token request to Seed 2.0 Pro is billed at $1 / $6 per 1M, not $0.50 / $3. BytePlus bills the tier by the prompt length of each request. Seed 2.1 Turbo has no step: $0.50 / $2.50 all the way to 256K. If your prompts are long, that makes Turbo cheaper than Pro on both input and output.

Seed models retiring on Nov 11, 2026

BytePlus says these models retire on November 11, 2026, per its pricing page. Seed 1.6, Seed 1.6 Flash and Seed 1.8 move to the 2.0 generation, and older dated builds of Seed 2.0 Lite and Mini move to the newer builds at the same list price. Change the model ID in your code before then.

Retiring model ID	BytePlus' recommended replacement
`seed-2-0-lite-260228`	`seed-2-0-lite-260428`
`seed-2-0-mini-260215`	`seed-2-0-mini-260428`
`seed-1-8-251228`	`seed-2-0-lite-260428`
`seed-1-6-250915`	`seed-2-0-lite-260428`
`seed-1-6-flash-250715`	`seed-2-0-mini-260428`
`deepseek-v3-2-251201`	`deepseek-v4-flash-ga-260731`
`glm-4-7-251222`	`dola-seed-2-1-turbo-260628`

Seed 2.1 Turbo rate limits

Seed 2.1 Turbo has a much lower ceiling than the 2.0 models: 500 requests and 1M tokens per minute, against 30,000 requests and 1.5M tokens per minute for Seed 2.0 Pro, Lite and Mini, per BytePlus' model list. If you fan out many small calls, test Turbo under load before you move a production job.

Is ByteDance Seed the same as Doubao?

Same family, different storefront. In mainland China ByteDance sells it through Volcano Engine as Doubao, with model IDs like doubao-seed-2-1-turbo, priced in yuan. Internationally, BytePlus ModelArk sells it as Seed, with IDs like dola-seed-2-1-turbo, in US dollars. ByteDance's Seed 2.1 announcement covers a Pro and a Turbo model; only Turbo has a BytePlus price so far. This page covers the BytePlus prices.

What a month costs

List price, no discounts, our math. Medium is 20M input and 5M output tokens a month; heavy is 200M and 50M.

Model	Medium month	Heavy month
Seed 2.0 Mini	$4.00	$40.00
Seed 2.0 Lite	$15.00	$150
Seed 2.1 Turbo	$22.50	$225
Seed 2.0 Pro	$25.00	$250
Seed 2.0 Code Preview	$25.00	$250

Seed against the models people compare it with

Seed 2.1 Turbo vs DeepSeek V4.1 Flash: $1 vs $0.52 blended. DeepSeek is cheaper and has a 1M window; Turbo has no peak-hour pricing.
Seed 2.1 Turbo vs Gemini 3.8 Flash: $1 vs $1.50 blended.
Seed 2.0 Pro vs Gemini 4 Argon: 3.6x cheaper blended.
All Chinese labs side by side: best Chinese LLM APIs.

ByteDance Seed API pricing explained