Deepseek Token API Pricing, Top-up, Deals & Updates | Article

Deepseek Token API pricing. DeepSeek-V4-Flash, DeepSeek-V4-Pro and 1 more models. OpenAI-compatible API, Anthropic-compatible API, Claude Code and 3 more tools integrations. Includes cache tiers and rate limits. API key setup, usage quotas, and official entry—is it worth it?

41
Platforms
395+
Models
140+
IDE Tools
3
Deepseek Token API models
¥1.5–3 / ¥4.5–9
Deepseek Token API entry price
View
Vendor details

DeepSeek API Latest Updates

DeepSeek V4-Flash Official Release: Beats V4-Pro Preview on All Agent Benchmarks, MIT Open Weights

Published Updated
ModelDeepSeekV4-FlashModel ReleaseAgentAPI UpdateOpen SourceCodex
Summary

DeepSeek V4-Flash is now officially released, with the API in public beta. Through re-post-training alone, it beats the V4-Pro preview across all 9 agent benchmarks. Adds native Responses API and Codex support, MIT open weights released same day. Pricing unchanged at $0.14 input / $0.28 output per MTok (cache-miss), $0.0028/MTok cache-hit, 2500 concurrency.

DeepSeek-V4-Flash is officially out of preview. The headline that really matters: a model with only 13B active parameters just beat the 49B-active V4-Pro preview on all nine agent benchmarks — purely through post-training, architecture untouched.

On July 31, 2026, DeepSeek-V4-Flash graduated to official status (DeepSeek-V4-Flash-0731), with public API access opening same day (source). Weights are MIT-licensed. Pricing is unchanged.

Same Architecture, Way Better Capability

Same CSA + HCA design. Same 284B total / 13B active parameters. Just one round of re-post-training. You can literally drop in the new weights — zero inference code changes.

Metric DeepSeek-V4-Flash-0731
Total Parameters 284B
Active Parameters 13B
Architecture CSA + HCA
Context Length 1M
Max Output 384K
Precision FP4 + FP8 Mixed
New DSpark speculative decoding
License MIT

DSpark is a new acceleration module — it speeds up decoding without touching anything in the output path.

Agent Benchmarks: 13B Bodies 49B

This isn't incremental. The post-training jump across all nine benchmarks is a full tier up (source):

Benchmark V4-Flash-0731
Terminal Bench 2.1 82.7
NL2Repo 54.2
Cybergym 76.7
DeepSWE 54.4
Toolathlon verified 70.3
Agent Last Exam 25.2
Automation Bench (Public) 25.1
DSBench-FullStack 68.7
DSBench-Hard 59.6

DSBench is proprietary — use for directional comparison only, not cross-model absolute scores.

What's New

Responses API

Native support for OpenAI's Responses API — the stateful replacement for Chat Completions, built for agentic workloads. V4-Pro gets this in early August.

Codex Adaptation

Specifically tuned for Codex, the agent framework that powers Claude Code. If you're in the Claude Code + DeepSeek API camp, this is a real quality-of-life improvement. See the official config guide.

DSpark Speculative Decoding

Pure throughput play — doesn't change model output, doesn't add parameters, just makes tokens move faster.

Pricing and Concurrency

Exactly the same as preview:

Item Price (USD) Price (CNY)
Cache-hit input $0.0028 / MTok ¥0.02 / MTok
Cache-miss input $0.14 / MTok ¥1 / MTok
Output $0.28 / MTok ¥2 / MTok

Concurrency 2500. Thinking/non-thinking dual mode, JSON Output, Tool Calls — all there.

Worth noting: the official pricing page also carries a price-hike warning. We covered this in a separate piece, but the short version: get your caching strategy dialed in now.

Open Weights

MIT license, available on both HuggingFace and ModelScope:

Base and Instruct variants. Fine-tune with Base, deploy with Instruct.

What to Watch

  1. Flash only this time — the Pro API, app, and web models are untouched.
  2. V4-Pro official is "coming soon," per DeepSeek's own words.
  3. deepseek-chat and deepseek-reasoner are fully retired (July 24, 15:59 UTC). They now route to v4-flash's non-thinking/thinking modes.
  4. Don't sleep on the price-hike warning — calc your monthly spend now and lock in your caching strategy.

Sources and verification

Last verified

Deepseek Token API FAQ

What actually changed between the official DeepSeek-V4-Flash release and the preview, and how did 13B active parameters beat V4-Pro with re-post-training alone?

The official release (DeepSeek-V4-Flash-0731) uses the exact same architecture and parameter count (284B total / 13B active) as the preview. It was only re-post-trained. That one step was enough to beat the 49B-active V4-Pro preview across all 9 agent benchmarks. The release also added native Responses API support, Codex adaptation, and a DSpark speculative-decoding module for faster inference. API calling is unchanged — model name stays deepseek-v4-flash.

What are the specific benchmark scores for the official DeepSeek-V4-Flash across all 9 agentic tests and the test configuration used?

Terminal Bench 2.1 (82.7), NL2Repo (54.2), Cybergym (76.7), DeepSWE (54.4), Toolathlon verified (70.3), Agent Last Exam (25.2), Automation Bench Public (25.1), DSBench-FullStack (68.7), DSBench-Hard (59.6). All 9 crushed the V4-Pro preview. Test config: DeepSeek Harness minimal mode, max effort level, top_p=0.95, temperature=1.0. DSBench-FullStack and DSBench-Hard are proprietary test sets — useful for relative comparison only.

What does the V4-Flash official release's native Responses API support and Codex adaptation mean for Claude Code integration?

Responses API is OpenAI's next-gen stateful API — way better for multi-turn agent interactions than the legacy Chat Completions endpoint. V4-Flash supports it natively with no config changes needed. The Codex adaptation specifically optimizes Claude Code's agent framework. If you're running Claude Code with DeepSeek API, this update is a noticeable upgrade. V4-Pro's Responses API support lands in early August.

When is the DeepSeek-V4-Pro official release coming after V4-Flash, and has the API pricing changed?

DeepSeek says the V4-Pro official release 'will follow soon.' This update only touches Flash — the Pro API, app, and web models are all unchanged. Pricing is identical to preview: cache-miss input $0.14/MTok, output $0.28/MTok, cache-hit input $0.0028/MTok, concurrency 2500. One thing to note: the official pricing page also has a warning about a significant price hike coming soon — plan ahead.

Where to download the MIT-licensed DeepSeek-V4-Flash open weights and what's the difference between Base and Instruct variants?

MIT license, released July 31 alongside the API. HuggingFace: deepseek-ai/DeepSeek-V4-Flash-0731, ModelScope: deepseek-ai/DeepSeek-V4-Flash-0731. Both Base and Instruct variants available in FP4+FP8 mixed precision. Base is for fine-tuning and post-training; Instruct is a chat-tuned checkpoint — ready to deploy with vLLM or SGLang. If you're just deploying, grab Instruct.
AI Token Plan

Global AI Token Comparison

AI Token Plan is not just a collection of vendor links. It places 41+ domestic and international AI platforms into one comparison framework, covering 395+ model entries plus common plan types such as Token Plans, Coding Plans, IDE Tools and LLM APIs.

When you need to compare AI subscription pricing, coding plan quota, API usage cost or official deal entry points, AI Token Plan brings official prices, plan tiers, usage rules, model capabilities and tool integrations into one place, reducing the need to check multiple vendor sites manually.

Data is continuously organized as vendor pricing pages, plan pages and product documentation change, making the homepage a pricing comparison entry point while detail pages explain whether each vendor plan fits individual developers, team purchasing or long-term API usage.

Prices come from each platform's official site and may change at any time; the official price prevails.

© 2026 AI Token Plan · All rights reserved · First published June 18, 2026 · 64 days running · Sitemap