DeepSeek is going to raise its API prices, and it will not tell you by how much or when. That is the entire content of the notice. Every run-rate you built on its price floor is now an estimate with no error bars.
Here is the part the wire reports missed. Because the weights are open, you do not have to wait for the number. Other companies already serve the identical model and already publish what it costs them to do it. That is your ceiling, and it is worse than "3x to 4x" — on cache hits, which is where agentic workloads live, it is 40x.
What DeepSeek Actually Posted on August 6
DeepSeek published a warning, not a price. Its official API pricing page now carries the line: "We plan to raise the overall pricing for DeepSeek API services in the near future, with a significant increase expected," followed by "The specific pricing plan will be subject to official notice."
No magnitude. No effective date. No stated reason. TechNode reported it on 6 August 2026, and The Next Web's read is the obvious one — the low prices drew a surge of users, and serving them strains the compute behind the service. The South China Morning Post framed it as a company struggling to hold an aggressively low price against fierce competition. Nobody has a number, because DeepSeek has not published one.
The current rates are still posted, which matters — the baseline that is about to move is documented today. DeepSeek V4-Flash is $0.14 per million input tokens on a cache miss, $0.28 output, and $0.0028 on a cache hit. V4-Pro is $0.435 in, $0.87 out, $0.003625 on a cache hit.
This is also the second pricing change in about five weeks. On 1 July DeepSeek announced peak-hour surcharging — every billing item at 2x during 9am–noon and 2pm–6pm Beijing time, with the stated goal of "better distribution of resources." That surcharge has not reached the pricing page yet — the posted rates are still flat, with no time-of-day tiers. But a vendor that announces time-of-day surcharging and then warns of an overall increase inside the same quarter is not making a one-time adjustment. It is repricing.
Open Weights Mean the Ceiling Is Public
You can price the worst case today, because the model DeepSeek sells is the same model anyone can download. DeepSeek-V4-Pro on Hugging Face is MIT-licensed — 1.6 trillion total parameters, 49 billion activated, 1M context. V4-Flash-0731 is MIT-licensed too, at 304B total parameters. MIT means commercial use, modification and redistribution with no restriction and no negotiation.
So look at what companies that must actually cover their own compute charge for it. Fireworks AI's serverless price list puts DeepSeek V4-Pro at $1.74 input / $0.145 cached / $3.48 output per million tokens. DeepInfra lists the same model at $1.30 / $0.10 / $2.60.
Against DeepSeek's first-party $0.435 / $0.003625 / $0.87, that is 3x at DeepInfra and exactly 4x at Fireworks on both input and output.
That number is your model, not a guess. DeepSeek's own eventual price is unknowable this week. The independent market price for serving these weights is published, competitive, and sitting at 3-4x. If your business case survives a 4x on V4-Pro, you have nothing to decide. If it does not, you have a decision and roughly one quarter to make it.
One caveat on shopping the marketplaces: OpenRouter currently lists V4-Pro at $0.338 / $0.676, below DeepSeek's own posted rate, with an 80% discount flagged on the listing. A promotional routed rate is not a contracted rate. Do not build a run-rate on one.
The Cache-Hit Gap Is Where the Damage Is
The 4x is the headline number and it is the smaller problem. The cache-hit price is where a repricing actually lands, and there the gap is an order of magnitude wider.
Compare cache-hit input, per million tokens:
| Model | DeepSeek first-party | Fireworks | Multiple |
|---|---|---|---|
| V4-Pro | $0.003625 | $0.145 | 40x |
| V4-Flash | $0.0028 | $0.028 | 10x |
DeepSeek prices a cache hit at roughly 1/120th of a cache miss on V4-Pro. Fireworks prices one at 1/12th. Those are different economic models, not different discounts.
This matters more than it looks because agentic workloads are cache-heavy by construction. A multi-turn agent replays a large, stable prefix — system prompt, tool definitions, retrieved context — on every single step, and that prefix is a cache hit every time after the first. We have written before about how prompt caching quietly restructured agent economics and made per-token list prices a poor proxy for what you actually pay. This is the same trap, pointed the other way.
Run the number for your own traffic. If your cache-hit ratio is 70%, a move from DeepSeek to Fireworks on V4-Pro is not a flat 4x on the input side — blended across the mix it is about 4.7x. At a 90% hit ratio it is 6.5x, and a prefix that is almost entirely cached converges on the full 40x. The lower your cache-miss share, the worse the port looks — which is the exact opposite of the intuition most FinOps models encode.
V4-Flash Is Already Hedged. V4-Pro Is Not.
Split your workloads by model before you do anything else, because the exposure is wildly asymmetric.
On posted cache-miss and output rates, Fireworks serves V4-Flash and V4-Flash-0731 at $0.14 / $0.28 — identical to DeepSeek's first-party rate, to the cent. Flash is a 304B model with a small activated parameter count; it is cheap enough to serve that a commercial host matches DeepSeek without subsidy. If your traffic is Flash and cache-miss-dominated, your hedge is already priced at par and the port is a base-URL change.
V4-Pro is the opposite. 1.6 trillion parameters with 49B activated is expensive to serve, and every independent host prices it well above DeepSeek: 3x at DeepInfra, 4x at Fireworks and Baseten, and about 5x at Together AI, which lists $2.10 input / $0.20 cached / $4.40 output. The 3-4x ceiling is what you pay if you shop; it is not the whole spread. That is the workload carrying your risk. It is also, in most shops, the workload doing the reasoning and the agent orchestration — the expensive, hard-to-substitute part.
The practical read: your Flash spend is not the problem, and your Pro spend is the whole problem. If you have not separated them in your cost reporting, you cannot size this at all.
Self-Hosting Is Not the Hedge Most People Think
Renting your own GPUs is a worse deal than the API for almost everyone, and the arithmetic is short enough to check in your head.
The V4-Flash-0731 model card documents deployment on a single 4×GB300 node under vLLM or SGLang. Fireworks publishes on-demand GPU rates of $12.00/hour for a B300 288GB — so call that node $48/hour. At DeepSeek's posted V4-Flash output price of $0.28 per million tokens, $48 buys about 171 million output tokens.
To break even on output tokens alone, four GPUs would have to sustain roughly 47,600 output tokens per second, every second, for the entire hour. No four-GPU node does that. Counting the input spend you would also stop paying lowers the bar — at a heavy 10:1 input-to-output mix on cache misses, to roughly 8,000 output tokens per second — which a four-GPU node is still unlikely to clear, and which collapses back toward 47,600 the moment your prefix is cached, because a cached input token costs $0.0028. Even on the cheapest reserved capacity — GB300 listings start around $5.60/hour per GPU on a one-year commitment — you still need over 22,000 tokens per second sustained, and the same listing page notes most GB300 capacity is out of stock or waitlisted. A hedge you cannot provision is not a hedge.
Self-hosting becomes rational at genuine scale, with high sustained utilisation, or when a compliance requirement makes the API unusable at any price. This is the same wall we hit when AMD's inference discount turned out to depend on a GPU nobody could rent, and the same reason NVIDIA's free 550B Nemotron did not make self-hosting free. Open weights remove the licence cost. They do not remove the GPUs.
For most readers the real hedge is a third-party host of the same weights. Price Fireworks AI, Together AI, Baseten and DeepInfra against your actual token mix, not against a list price. Check availability before you plan around a host: Amazon Bedrock currently lists DeepSeek V3.1, V3.2 and R1 and not V4, so it is not a like-for-like exit today.
The Port That Also Closes a Compliance Gap
Moving off DeepSeek's first-party API fixes a problem your security team probably already raised. DeepSeek's privacy policy states plainly: "To provide you with our services, we directly collect, process and store your Personal Data in People's Republic of China." There is no region selector.
For a regulated workload that is not a preference, it is a blocker — and for those teams the first-party endpoint was never available in the first place, which makes a Western host of the same weights the only version of DeepSeek they could ever have run. We covered the supply-chain and jurisdiction side of this when Chinese models crossed 46% of routed enterprise token volume, and the three risks that survive the 90% cost saving have not changed.
So the hedge you are being forced into for cost reasons is the one your governance function wanted anyway. That is unusual. Take the win — but book the cost honestly, because you are paying 3-4x for it and calling it compliance does not make it free.
The Steel Man: This Repricing Was Always Coming
DeepSeek's prices were not a market clearing price, and pretending otherwise was the modelling error.
Look at the volume-versus-spend split. Vercel's July 2026 AI Gateway index — one gateway's routed traffic, reporting June data, not a market-wide measure — put DeepSeek at 22.6% of gateway token volume, third behind Anthropic and Google, while its share of spend was negligible. Open-weight models overall took 29% of volume against just under 4% of spend. The June index, covering May, had DeepSeek at 17% of tokens and roughly 1% of spend, against Anthropic's 32% of tokens and 65% of spend.
Treat that as one well-instrumented sample rather than the market — Vercel's gateway skews toward its own developer base, and no comparable industry-wide split is published. But a provider carrying a fifth of a major gateway's tokens for approximately none of its revenue is running a land-grab, not a business model. Nobody sustains that indefinitely. The honest version of the criticism is that buyers who treated $0.435 as a durable input to a three-year plan made the mistake, not DeepSeek.
There is a wider consequence too. That same July index notes closed-weight frontier prices rose about 12% per token, and the industry's average price per token stayed flat only because cheap open-weight volume grew to offset it. DeepSeek's volume is a meaningful part of that offset. If DeepSeek reprices toward what independent hosts charge, the flat line stops being flat — and every multi-model routing strategy built on a cheap fallback tier gets less headroom at once.
Do This Before the Notice Lands
You have leverage now that you will not have on the day the number is published. Use it.
This Week:
- Split your DeepSeek spend by model and by cache-hit ratio. Pro versus Flash, hit versus miss, per workload. Without those four numbers you cannot size the exposure, and every step below depends on them.
- Re-run your run-rate at 4x on V4-Pro cache misses and 40x on V4-Pro cache hits. Not as a scenario deck — as the new base case, until DeepSeek publishes something better. Take that number to whoever signed off on the original business case, this week, before it arrives as a surprise.
- Send one workload's traffic to a second host today. Fireworks, DeepInfra, Together or Baseten. Not a proof of concept — a real slice of production traffic, so you learn what actually breaks before you need the exit.
This Month:
- Confirm the weights are actually identical. Third-party hosts pin a checkpoint; DeepSeek's first-party endpoint has moved under people before. That is exactly how a silent post-training swap invalidated agent evals without a version bump. Pin the checkpoint by name in your config and re-run your eval suite against each candidate host.
- Price the cache economics, not the sticker. Ask each host directly: what is the cache-hit rate, what is the TTL, and is it per-tenant. A host with no prefix caching at any price is disqualified for agentic traffic regardless of its input rate.
- Put a repricing clause in anything you sign. Any host, any term. Notice period, cap on increase, right to terminate without penalty. This is the same class of exposure as an acquirer repricing your inference credits mid-contract — you get the protection at signature or you never get it.
Before Renewal:
- Set a hard trigger. Decide now, in writing, at what blended cost per million tokens you move each workload, and who executes it. A threshold agreed under no time pressure is worth more than a review scheduled for after the invoice.
The Bottom Line
The valuable thing here is not that DeepSeek is raising prices. Vendors raise prices. The valuable thing is that this is a frontier-tier repricing where the buyer can compute the worst case before the vendor announces it — because the weights are MIT-licensed and the substitutes publish their rates.
That is what open weights actually buy an enterprise. Not free inference; the GPU math kills that for almost everyone. What they buy is a priced, published ceiling on how badly a single vendor can reprice you — 3-4x on posted rates, 40x on cache hits, knowable today. Every closed-model dependency in your stack has the same exposure and no such ceiling. When OpenAI or Anthropic sends the next pricing email, there is no second seller of those weights to check against, and the number you get is the number you pay.
DeepSeek did not give you a price. It gave you a deadline.
Continue Reading
- DeepSeek Swapped the Model. Your Eval Didn't Notice.
- Your AI Router Is Trading a 10x Discount for a 2.5x One
- AMD's Inference Discount Depends on a GPU You Can't Rent
- 46% of Your AI Now Runs on Chinese Models
- Claude vs GPT vs Gemini: Stop Comparing Per-Token Prices
- Free 550B Model: NVIDIA Ends Self-Hosted AI Quality Gap
- Bending Spoons Bought Airtable. Reprice Before It Closes.
- Chinese AI: 90% Cheaper—3 Risks Every CIO Can't Ignore