DeepSeek V3
DeepSeek · Released December 2024
Input Price
$0.28
per 1M tokens
Output Price
$0.42
per 1M tokens
Context Window
128K
tokens
Speed Score
85
out of 100
Benchmarks
Avg 88/100Pricing
Input
Prompt / context tokens
$0.28/ 1M
Output
Generated tokens
$0.42/ 1M
Best For
DeepSeek V3 in context
On a typical 70% input / 30% output mix, DeepSeek V3 blends to $0.32 per million tokens. That puts it above 16% of the 32 models tracked here, and 5 other models in the same budget tier undercut it. Tier is a capability label, not a price band — the spread inside one tier is often wider than the gap between tiers, which is why picking by tier alone tends to overspend.
Input runs $0.28 per million and output $0.42 — a 1.5x ratio. That ratio matters more than either number on its own: a summarisation or classification workload reads far more than it writes and will track the input price, while a code-generation or long-form writing workload inverts that and will track the output price. Compare models on the ratio your own traffic actually has, not on the input price alone.
The benchmark profile is unusually flat — coding leads at 91/100 but only 6 points separate the strongest category from the weakest. That makes DeepSeek V3 a safe default for mixed workloads where you cannot predict in advance which capability a request will lean on, and a poor choice if you need a specialist.
The context window is 128K tokens and the throughput score is 85/100. Context only earns its keep if you fill it, and filling it is also what makes a request expensive — a large window is an option, not a discount. For a budget-tier model, judge the speed score against what you are doing: it decides everything for interactive chat and almost nothing for overnight batch work.
Nothing cheaper in the table matches DeepSeek V3 on combined coding and reasoning, so its $0.32 blended price is buying capability you cannot get for less right now. That is the case for paying it — and it is worth re-checking, because this is exactly the position that a new release takes away.
The closest comparison is GPT-5.6 Luna from OpenAI, at $0.50 per million against $0.32. 0 models in this tier score higher on combined coding and reasoning, which is the useful framing: the question is rarely whether a model is good, it is whether it is the cheapest thing that is good enough for the specific work you are sending it.
What it costs in production
At 10M + 2M tokens a month — a realistic mid-size production workload — DeepSeek V3 runs about $3.64. There is no batch discount on this model, so that figure is the floor — the usual lever of shifting background work to a cheaper asynchronous tier is not available here.
Price History
No price changes since launch
Strengths
- Best-in-class open source coding performance
- Ultra-low pricing — $0.14 per million input tokens
- Strong mathematical reasoning
- Good Chinese language support
Avoid For
- Vision tasks (text-only model)
- GDPR-strict deployments
- Real-time low-latency needs