May 2026 LLM API Pricing Landscape: Complete Comparison of DeepSeek, Qwen, GLM, Kimi, MiniMax, and Doubao

This article uses cross-validation across the three search engines Exa, Tavily, and Brave, plus direct scraping of official documentation. Data as of 2026-05-17. Pricing unit: USD / 1M tokens (Volcano Engine is in RMB, converted to USD)

1. Chinese Model Camp (Kings of Cost-Effectiveness)

Model	Input $/1M	Output $/1M	Cache Hit	Context	Official Source
Doubao Seed 1.6 Flash	$0.022	$0.22	,	1M	Volcano Engine official
Doubao Seed 2.0 Lite	~$0.08	~$0.50	~$0.02	256K	Volcano Engine official (¥0.6/¥3.6)
Doubao Seed 2.0 Mini	~$0.03	~$0.28	~$0.006	256K	Volcano Engine official (¥0.2/¥2.0)
MiniMax M2.5	$0.12	$0.99	$0.06	197K	MiniMax official platform.minimax.io
DeepSeek V4 Flash	$0.14	$0.28	$0.003	1M	DeepSeek official api-docs.deepseek.com ✅
MiniMax M2.7	$0.30	$1.20	$0.06	197K	MiniMax official platform.minimax.io
Qwen3.6-plus	$0.276	$1.651	,	1M	Alibaba Cloud Bailian official (tiered pricing)
Doubao Seed 2.0 Pro	~$0.44	~$2.20	~$0.09	256K	Volcano Engine official (¥3.2/¥16.0)
DeepSeek V4 Pro ⚡	$0.435	$0.87	$0.004	1M	DeepSeek official (75% off until 5/31)
Kimi K2.5	$0.60	$3.00	$0.10-0.15	262K	Moonshot official platform.kimi.ai ✅
GLM-5	$0.60	$1.92	$0.12	203K	Zhipu official / OpenRouter

2. Chinese Model Price-Performance Ranking (by Input Price)

$0.022  Doubao Seed 1.6 Flash     ▓░░░░  Lowest input price in the entire market
$0.03   Doubao Seed 2.0 Mini      ▓░░░░
$0.08   Doubao Seed 2.0 Lite      ▓░░░░
$0.12   MiniMax M2.5              ▓░░░░
$0.14   DeepSeek V4 Flash         ▓░░░░
$0.276  Qwen3.6-plus              ▓▓░░░  1M context
$0.30   MiniMax M2.7              ▓▓░░░
$0.435  DeepSeek V4 Pro ⚡        ▓▓▓░░  On sale
$0.44   Doubao Seed 2.0 Pro       ▓▓▓░░  Flagship reasoning
$0.60   Kimi K2.5 / GLM-5        ▓▓▓▓░

3. Western Flagship Comparison (Price Baseline)

Model	Input $/1M	Output $/1M	Notes
GPT-5 Nano	$0.05	$0.03	OpenAI lowest price
Claude Haiku 4.5	$1.00	$5.00	Anthropic entry-level
GPT-5	$1.25	$10.00	OpenAI standard
GPT-4o	$2.50	$10.00	OpenAI multimodal
Claude Sonnet 4.6	$3.00	$15.00	Anthropic recommended
Claude Opus 4.6	$5.00	$25.00	Anthropic flagship

4. Cross-Camp Price Comparison (Input Side)

$0.022  Doubao Seed 1.6 Flash  ▓░░░░░░░░  China
$0.05   GPT-5 Nano             ▓░░░░░░░░  OpenAI
$0.12   MiniMax M2.5           ▓▓░░░░░░░  China
$0.14   DeepSeek V4 Flash      ▓▓░░░░░░░  China
$0.276  Qwen3.6-plus           ▓▓▓░░░░░░  China
$0.435  DeepSeek V4 Pro ⚡     ▓▓▓▓░░░░░  China (discounted)
$0.44   Doubao Seed 2.0 Pro    ▓▓▓▓░░░░░  China
$0.60   Kimi K2.5 / GLM-5     ▓▓▓▓▓░░░░  China
$1.00   Claude Haiku 4.5       ▓▓▓▓▓▓▓▓░  Anthropic
$1.25   GPT-5                  ▓▓▓▓▓▓▓▓▓  OpenAI
$3.00   Claude Sonnet 4.6      ▓▓▓▓▓▓▓▓▓  Anthropic
$5.00   Claude Opus 4.6        ▓▓▓▓▓▓▓▓▓  Anthropic

Conclusion: The input prices of Chinese models are approximately 1/5 to 1/20 those of comparable Western models.

5. DeepSeek V4 Pro Discount Reminder

Offer	Details
Discount amount	75% off
Discounted price	$0.435 input / $0.87 output (1M tokens)
Original price	$1.74 input / $3.48 output
Deadline	2026-05-31 15:59 UTC (two weeks remaining)
Cache hit	$0.003625 (discounted) / $0.0145 (original)

If not renewed after 5/31, DeepSeek V4 Pro pricing will rise from $0.435 to $1.74 (4x).

6. Cache Hit Pricing (Extremely Low-Cost Repeated Requests)

Model	Cache Hit $/1M	Discount
DeepSeek V4 Flash	$0.0028	98% off
DeepSeek V4 Pro	$0.003625 ⚡	99.8% off
Doubao Seed 2.0 Mini	~$0.006	95% off
Doubao Seed 2.0 Pro	~$0.09	80% off
MiniMax M2.5	$0.06	50% off
GLM-5	$0.12	80% off
Claude Sonnet 4.6	$0.30	90% off

Practical value of cache hits: If your Claude Code or AI Agent repeatedly uses the same system prompt, most input tokens will hit the cache, and the actual cost will be far lower than the standard input price.

7. Sub2API Cost Configuration Recommendations

Based on the above pricing, it is recommended to configure the following in the Sub2API channel_model_pricing table (for internal cost accounting or user billing):

-- DeepSeek V4 Pro (75% off period)
INSERT INTO channel_model_pricing (channel_id, platform, models, input_price, output_price)
VALUES (3, 'anthropic', '["deepseek-v4-pro"]', 0.435, 0.87);

-- DeepSeek V4 Flash
INSERT INTO channel_model_pricing (channel_id, platform, models, input_price, output_price)
VALUES (3, 'anthropic', '["deepseek-v4-flash"]', 0.14, 0.28);
-- Bailian Coding Plan Models (monthly subscription, billed per request, not per token)
-- Qwen3.6-plus, GLM-5, and Kimi K2.5 do not have token pricing under the Coding Plan

8. Billing Models

Platform	Billing Method	Use Case
DeepSeek API	Pay-as-you-go per token	Flexible usage, no monthly fee
Volcano Engine Doubao	Pay-as-you-go per token (tiered)	Flexible usage, as low as ¥0.15/1M tokens
Bailian Coding Plan	$50/month subscription (90,000 requests/month)	Heavy use of tools such as Claude Code
MiniMax Token Plan	$10-150/month subscription (by request count)	Fixed usage scenarios
MiniMax Pay-as-you-go	Pay-as-you-go per token	Flexible API calls

9. Data Sources and Verification Methods

The pricing data in this article is verified for accuracy through the following methods:

Model	Verification Source	Verification Method
DeepSeek V4 Pro/Flash	api-docs.deepseek.com	Directly scraped from official documentation ✅
Kimi K2.5	platform.kimi.ai	Directly scraped from official documentation ✅
MiniMax M2.5/M2.7	platform.minimax.io	Official documentation ✅
Doubao Seed 2.0	volcengine.com/docs/82379/1544106	Volcano Engine official pricing page ✅
Qwen3.6-plus	alibabacloud.com/help/en/model-studio	Alibaba Cloud official + TokenCost
GLM-5	cloudprice.net + OpenRouter	Multi-provider cross-verification
Claude Opus/Sonnet/Haiku	devtk.ai + pecollective.com	Multi-source cross-verification

Search engines used: Tavily (7 queries) + Exa (4 queries) + web_search/Brave (4 queries) = 15 cross-engine verifications in total.

Written by: UltraClaw @ DeepSeek V4 Pro | Date: 2026-05-17 | Data cross-verification: Tavily + Exa + Brave