DeepSeek V4 Pro 0813 Max vs Kimi K3 Low: which is cheaper?

Here is how DeepSeek V4 Pro 0813 (Max), built by DeepSeek, lines up against Kimi K3 (Low) from Kimi. Price separates them more than anything else; on a blended basis Kimi K3 Low costs 3 times as much as DeepSeek V4 Pro 0813 Max. What follows is built entirely from tracked data, so where the catalogue is silent, this page says so instead of guessing.

Running costs

Per million tokens, DeepSeek V4 Pro 0813 Max charges $1.32 for input and $3.96 for output, while Kimi K3 Low charges $3.00 and $15.00. For output tokens, Kimi K3 Low is 3.8 times the price of DeepSeek V4 Pro 0813 Max.

A reference job of 1 million input tokens and 200,000 output tokens costs $2.11 with DeepSeek V4 Pro 0813 Max and $6.00 with Kimi K3 Low, so DeepSeek V4 Pro 0813 Max saves $3.89. Scale it up to classifying 100,000 short messages at 150 input tokens and 5 output tokens each and the bill is $21.78 for DeepSeek V4 Pro 0813 Max against $52.50 for Kimi K3 Low. On a 3:1 blend of input to output tokens, that works out to $1.98 versus $6.00 per million.

Reasoning and overall scores

DeepSeek V4 Pro 0813 Max comes out ahead on the intelligence index, 23.1 to 17.4. It is a meaningful gap rather than a rounding difference.

For context, that places DeepSeek V4 Pro 0813 Max 72nd and Kimi K3 Low 86th out of 109 scored models still on offer.

For coding or maths work specifically, the catalogue has no dedicated index on either side yet.

Score per dollar

A blunt value measure, blended price over intelligence score, puts DeepSeek V4 Pro 0813 Max at $0.086 and Kimi K3 Low at $0.34 per point. With the higher score and the lower cost per point, DeepSeek V4 Pro 0813 Max is the better value per point on these numbers.

DeepSeek V4 Pro 0813 Max is the faster writer

Once a response is under way, DeepSeek V4 Pro 0813 Max produces roughly 67 tokens per second against 49 tokens per second for Kimi K3 Low. That makes DeepSeek V4 Pro 0813 Max roughly 1.4 times as fast at writing. Time to first token is 1.7 seconds for DeepSeek V4 Pro 0813 Max and 3.6 seconds for Kimi K3 Low. In an interactive app, DeepSeek V4 Pro 0813 Max will feel noticeably more responsive.

For a 1,500-token response, first token plus generation comes to roughly 24 seconds on DeepSeek V4 Pro 0813 Max and 34 seconds on Kimi K3 Low.

Generation gap

DeepSeek V4 Pro 0813 Max arrived on August 13, 2026, 28 days after Kimi K3 Low (July 16, 2026). That is close enough to count them as the same generation.