DeepSeek Code Plan
Three search intents, zero official plan menu
As of 2026-10-13 the official DeepSeek Models & Pricing page bills strictly per token from your topped-up or granted balance and lists only the two model names deepseek-flash and deepseek-v4-pro; nothing is sold as a Code Plan or Coding Plan subscription. Whether a monthly coding price menu found on other sites can back a quote is answered in the Caution section below.
Updated 2026-10-14
One key across Claude, GPT and Chinese models, billed per token.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Highlights
Official plan name
No Coding Plan on the pricing page.
Context length
The context length shown on the official pricing page.
How they charge
Prepaid tokens. Flash/Pro prices + 8-16 peak/off-peak.
The real “plan”
The DeepSeek Harness documentation has its agent keep a plan while it works, and setup asks for your own DeepSeek API key. That plan is a plan inside the tool, not a monthly subscription.
What the query is really asking
Three things usually sit behind this query: the 2026 roadmap of DeepSeek's coding models (the timeline below: V4 entered the API on 2026-04-24, the V4-Pro GA release came on 2026-08-13, V4.1-Flash was released on 2026-09-10), a monthly allowance for coding, and the plan kept by the DeepSeek Harness agent. As of 2026-10-13 the official pricing page names only deepseek-flash and deepseek-v4-pro, bills per token, and lists no monthly plan called Coding Plan or Code Plan.
Why the phrase keeps being searched
The newest entry in the official changelog is dated 2026-09-10: DeepSeek released the DeepSeek-V4.1-Flash model and wrote that API prices were reduced accordingly. Nothing was added up to the 2026-10-13 check, and no monthly coding plan was announced. Whether such a plan exists comes down to two official documents, the Models & Pricing page and the changelog, and this page carries only what those two show word for word.
Timeline
2026-04-24 V4-Pro / V4-Flash entered the API.
2026-05 Code Harness hiring: Model + Harness = Agent.
2026-07-31 0731, Harness preview in 2026-08, 0813 GA on 2026-08-13, peak/off-peak from 2026-08-16; deepseek-v4-flash retired on 2026-09-10.
Confirmed vs caution
Confirmed
What verifies word for word on 2026-10-13: prices are listed in units of per 1M tokens, billing follows the total number of input and output tokens, and the fees are deducted directly from your topped-up balance or granted balance. The pricing table carries only the two model names deepseek-flash and deepseek-v4-pro, with a 1M context length, and off-peak rates are half of the peak rates. DeepSeek V4 Pro continues after September 14, 2026 with the billing method unchanged, and the newest changelog entry is dated 2026-09-10. Neither the pricing page nor the changelog lists a Code Plan or Coding Plan SKU. The coding route the docs do give is your own API key: Claude Code is pointed at the DeepSeek Anthropic API, where names starting with claude-opus map to deepseek-v4-pro and those starting with claude-haiku or claude-sonnet map to deepseek-flash, and DeepSeek Harness also asks for your own DeepSeek API key.
Caution
As of 2026-10-13 the official offering exists in one form only, token-metered billing: neither the pricing page nor the changelog lists a monthly coding tier, and there is no official coding-subscription price menu to compare against. A DeepSeek monthly coding price table you come across elsewhere has no official source behind it and cannot serve as the basis for a quote.
How to choose
You want a monthly coding allowance
OpenCode Go (third party) or accept official prepaid tokens.
Want DeepSeek’s own agent product
DeepSeek Harness (preview), not a fictional Code Plan SKU.
How to use it
① First decide whether you want a monthly allowance or metered use: officially there is only token-metered billing, priced per 1M tokens, with fees deducted directly from your topped-up or granted balance. ② For metered use, follow the official docs with your own DeepSeek API key and one of the two model names on the price page, deepseek-flash or deepseek-v4-pro; off-peak rates are half of the peak rates. ③ If you want a coding agent that keeps its own plan, configure the DeepSeek Harness as its documentation describes, which again asks for your own DeepSeek API key.
On QCode
QCode is one API key used over a standard API interface, covering Claude, GPT as well as GLM, Kimi, DeepSeek and Qwen models, which is a different thing from a DeepSeek subscription. As of 2026-10-13 the DeepSeek row on /models lists deepseek-v4.1-flash, deepseek-v4-flash and deepseek-v4-pro. Billed per token; per-model rates are listed on /models.
FAQ
Is there a DeepSeek Max plan?
No, as of 2026-10-13 the pricing page bills only deepseek-flash and deepseek-v4-pro per token from a topped-up balance, and neither the pricing page nor the changelog lists a plan named DeepSeek Max.
Then how are the fees deducted?
Per token: prices are listed in units of per 1M tokens, billing follows the total number of input and output tokens, and the fees are deducted directly from your topped-up balance or granted balance.
Can a DeepSeek coding monthly price menu found elsewhere be quoted?
Not as a basis: as of 2026-10-13 neither the official pricing page nor the changelog lists a matching monthly tier, and there is no official price menu to compare such a list with.
Is Harness the Code Plan?
It’s the closest official product: an open harness, not a monthly SKU.
OpenCode Go a DeepSeek plan?
No. As of 2026-10-13 the official pricing table names only deepseek-flash and deepseek-v4-pro and bills per token, so OpenCode Go is not on that table.
Peak pricing vs a plan?
From 8-16, official API uses peak/off-peak instead of a coding subscription.
Sources
Sources: the DeepSeek Models & Pricing page, the changelog in English and Chinese, the official Claude Code integration guide and the DeepSeek Harness docs, all re-crawled and checked on 2026-10-13; the QCode /models list checked the same day. No unofficial price menu is used on this page.
DeepSeek has no monthly Coding Plan; you can still start working per token
Pick the route that matches what you actually need: your own API, a third-party subscription, or a single multi-model key.
Related
Harness + Cordis
The developer preview guide for DeepSeek Harness and Cordis.
OpenCode Go
OpenCode Go limits and price: Go at $10, Plus at $40.
DeepSeek V4 Flash 0731 historical record
That page covers the benchmarks, the price at the time and the retirement on 2026-09-10.
Crypto payment guide
Nine coins, 56 networks, zero platform fee for AI coding API credits.
Rechecked on 2026-10-13 against the official DeepSeek pricing page and changelog. Not affiliated with DeepSeek, and no unofficial price menu here; prices and whether a plan exists follow the official pages, and QCode model availability follows /models.