DeepSeek V4 Pro
The cheapest frontier-class model on the market, and not by a small margin. 1.6T MoE / 49B active, MIT-licensed weights on Hugging Face, 1M context, 384K max output. DeepSeek launched it at $1.74/$3.48, cut prices 75% as a promotion, then made the cut permanent on 31 May 2026 — so the standing rate is $0.435 input / $0.87 output per 1M tokens, with cache hits at $0.003625. On Artificial Analysis's blended 7:2:1 basis that works out to about $0.18 per 1M tokens against $0.90 for GLM-5.2 and $2.31 for Kimi K3; measured cost-per-task is roughly $0.04 versus $0.94 for K3, a 23x gap. Quality is genuinely frontier-adjacent rather than budget: 80.6% on SWE-bench Verified (matching Gemini 3.1 Pro) and 83.5 on MRCR 1M. One dollar buys about 1.15M output tokens here versus 67K on Kimi K3.
1M
tokens
33K
tokens
$0.435
per 1M tokens
$0.87
per 1M tokens
62
tokens/sec
Apr 2026
2026-04-24
$0.65
per 1M tokens
137.9
quality per $
Capabilities
Benchmarks
Pricing History
Compare With
Open Source: Licensed under MIT
About DeepSeek V4 Pro
DeepSeek V4 Pro is a open-source AI model by DeepSeek, released on April 24, 2026. It supports a context window of 1000K tokens and can generate up to 33K output tokens.
At $0.435 per million input tokens and $0.87 per million output tokens, its blended cost of $0.65/1M tokens makes it one of the most affordable models available. Its value score of 137.9 reflects the balance of quality and cost.
DeepSeek V4 Pro is available as an open-source model under the MIT license, meaning you can self-host it for predictable costs or use it through API providers like Swfte Connect.
Using DeepSeek V4 Pro with Swfte
Access DeepSeek V4 Pro through Swfte Connect, our unified LLM gateway. Connect gives you a single API for 50+ models, with automatic routing, cost optimization, and fallback handling. You can also try DeepSeek V4 Pro in our AI Playground before integrating.