DeepSeek V4 Flash vs DeepSeek V4 Pro

Specs side by side

Price per token

Prices exclude taxes and are charged from your prepaid USD balance. Both models work through the same key, in the OpenAI-compatible /v1/chat/completions endpoint. Failed generations cost nothing.

Start with DeepSeek V4 Flash: it is the smaller model of the same V4 family, and on Jantaris it costs several times less than V4 Pro on both input and output - exact rates and the cost of 1,000 replies are in the table above. Use DeepSeek V4 Pro where your own data shows the larger model gives clearly better results, or where your prompts, parsers and tests were built around it. Both have a 1M-token context, replies up to 384K tokens and open weights.

Key differences

• Price. DeepSeek V4 Flash costs $0.135 / $0.27 per 1M tokens, DeepSeek V4 Pro costs $0.54 / $1.22 per 1M tokens. Flash is cheaper on both input and output. The cost of 1,000 replies on each model is in the table above.
• Size. V4 Pro is a MoE model with 1.6T parameters, 49B of them active per token. Flash is the smaller model of the same family, meant for lower-cost work.
• Family. Both belong to DeepSeek V4. Pro was in preview from April 2026 and became generally available in August 2026.
• Thinking. Both think before answering by default and offer low, high and max reasoning effort. Thinking tokens are billed as output.
• Shared. 1,000,000 tokens of context, replies up to 384,000 tokens, open weights, tool calls, JSON mode and the same OpenAI-compatible endpoint.

When to pick DeepSeek V4 Flash

• Coding agents that read a large chunk of the repository on every step.
• Bulk classification, data extraction and summaries of long documents.
• Customer support chat and other high-volume workloads.
• Tests and prototypes with thousands of requests.

When to pick DeepSeek V4 Pro

• Harder tasks where the larger model does clearly better on your data.
• Production pipelines that run on V4 Pro and have regression tests built on its outputs.
• An A/B run of both models on your data before migrating.
• Long-context analysis, if Pro does better on your prompts.

Cost in practice

On both models output costs more than input, and with high reasoning effort that share grows, because reasoning is billed as output too. In my tokenizer test (published in Polish on promptowy.com), V4 Pro used only about 5% more tokens than OpenAI models on the same Polish text; V4 Flash was not part of that measurement, so I do not quote a number for it. A wider ranking of budget models is in Cheapest LLM models in the Jantaris catalog, and other currently available pairs are listed in model comparisons.

I start with V4 Flash, and only when a specific task type comes out clearly worse do I run it in parallel on V4 Pro. Both models sit behind the same key and balance, so switching is a one-line change to model.

FAQ

DeepSeek V4 Flash or V4 Pro: which is better?

V4 Pro is the larger model, but I have no independent head-to-head test of this pair. With a several-fold price gap, a trial on your own prompts decides: if Flash is good enough, there is no reason to pay for Pro.

Which is cheaper: V4 Pro or V4 Flash?

V4 Flash, on both input and output. Current rates: $0.135 / $0.27 per 1M tokens and $0.54 / $1.22 per 1M tokens, the cost of 1,000 replies is in the table above.

Will DeepSeek V4 Pro be shut down?

DeepSeek first announced it would redirect V4 Pro traffic to a newer model, then walked that back and kept V4 Pro available. Still, expect a migration at some point and keep the model ID in your config.

Does moving from V4 Pro to V4 Flash need code changes?

Change model from deepseek-v4-pro to deepseek-v4-flash. The endpoint, key and response format stay the same. Just check that the smaller model's answers are good enough for your tasks.

Do I need a DeepSeek account?

No. You call both models with your Jantaris key and pay from one prepaid balance, and Jantaris does not use your prompts or outputs to train models.