Curious for folks who have made the switch I’m considering: if I swapped Claude Code to DeepSeek API pricing, would I get more bang for my buck compared to the $100 Max plan I’m using now?
I only hit the 5 hour limit every few days and the weekly limit a day or two before it resets at the most aggressive. I wouldn’t expect my usage to increase dramatically, other than not being stopped by limits.
I’m still apprehensive about shipping all my stuff off to a lab under an adversarial government (to the US), so not just looking at this from a pure cost basis, but my question is from the cost lens at the moment.
Depends on what you mean by 'bang for buck'. The open weights aren't better than openai/claude. But they are much cheaper and the limits are much higher, so you get more work out of it for less money. Every subscription provider out there provides better money-per-limit value than Anthropic (other than GitHub, who are by far the most embarrassingly overpriced and limited provider). (https://codeberg.org/mutablecc/calculate-ai-cost/src/branch/...)
> I’m still apprehensive about shipping all my stuff off to a lab under an adversarial government (to the US)
Do you mean you don't want to use the models created by a non-US lab? In that case, yes you're stuck with US models, but there's a half dozen big labs in the US. If you meant just where your inference is done, there are providers in 12 different countries through OpenRouter, including the US. Several subscription providers host in multiple countries. There's a lot of choices.
If you worry about sending your data off for inference, Fireworks is one of the companies serving open models with solid performance and compliance/zero data retention sorted out. OpenCode supports them and many others. Cursor uses them. They don't have the super-cheap cache reads deal that DeepSeek's own endpoint does, but are still well below Anthropic API rates. (Though crucially you're not paying API rates now!)
DeepSeek and Xiaomi's deals on cache reads go with their models' latest gens making caching cheaper (using less space for KVs). No open-model inference provider has decided to match the pricing. I'm sure that says something about how inference pricing works, but not completely sure what.
Agree with others that top open models aren't on the frontier, and I would expect differences doing big-picture planning or anywhere you're only giving broad brushstrokes and looking for a lot to be guessed. But they do seem fine at coding from a a concrete plan! No experience in huge codebases because I only use them outside work, but they seem good enough about gathering info before they dive in that I'd expect them to grep around as they need.
An annoying caveat: individual subscription plans, used heavily, are much cheaper than the API -- see https://she-llac.com/claude-limits -- which complicates any argument about cost. I still think open models are worth playing with. They're one of the things that let us treat this as a technology rather than just as the product offerings of one of a few companies.
My advice -- give it a try. Chuck $5 into deepseek.com , and use this config (put it in a shell script, run ' . ./deepseek-claude.sh ', then just run claude as normal.
I started by using it for some bigger reading jobs, particularly when I was near limit. Honestly, it's not quite as good, but it's much cheaper, and means I can carry on working. I also find sometimes it's good to ask claude and deepseek to consider code, how to polish, it see what they both say.I’m using Claude with a $100/month subscription. I’m playing around with using Opus as the Architect, Sonnet as the implementer/engineer and Deepseek-pro as the deep reviewer, and tester. It’s been quite good as I expected. If my usage pattern holds up, I would downgrade my subscription to the $20/month one and toss more money to Deepseek.
Repo reference here: https://github.com/aravindhsampath/agentic-template
Much more bang per dollar, yes. Somewhat less bang per hour.
As usual, different models get stuck on different things. I run DeepSeek v4 API for most of my Cursor experimentation / poking around / proof of concept stuff, but I trust it less than OpenAI/Claude for writing production code. Sometimes DeepSeek is great for debugging, planning, etc. Sometimes it gets stuck or outputs low quality. That's true of OpenAI and Anthropic models as well though.
Overall, DeepSeek seems serviceable but a rung below Opus 4.8 and GPT 5.5. I run them all on maximum thinking settings.
Deepseek cost/performance is incredible. That said, I still feel like for agentic coding we haven't plateaued (I slightly prefer GPT 5.5 to Claude for complex stuff, to be honest), and so the extra price is absolutely worth it to push you over the 'impossible' to 'feasible' bar on complex tasks. Once you're in a domain that Deepseek can handle though that requires volume, I would almost always default to it now.
For evals in particular (tuning workflows that agents are using), effectively not having to worry about price is an incredible multiplier - getting statistical significant signal is not cheap otherwise.
I've found myself liking opencode for workflows because i can plug GPT models into it, so i tossed 5$ at deepseek api and just toggle back and forth what my opencode.jsonc file is running model wise for my agents. I havent tried anything crazy yet with it, but its nailed all the tasks i felt were overall too simple to waste gpt usage on.
Hardest stuff i threw at it... i did like a set of 3 each for claude/gpt/ds, it was all pretty steady across all providers. I think claude won but it could have just been it rng'd into the 3 easier tasks, they are all similar tasks but not identical, these aren't like benchmark tasks just a steady flow of annoying html/json/regex type stuff. Almost always they need a second pass regardless of what model i throw at it, just to tighten up some loose ends, and it fit right into what my current expectation was of gpt 5.5 and opus 4.6.
I used ~16,000,000 input tokens yesterday on v4 pro, ~15,000,000 were cache hits, and I spent $0.47. Output tokens were negligible. However that's with Zed's harness, I'm not sure what you would get with Claude Code.
It's maybe not quite as knowledgeable as the most expensive American models and maybe makes more mistakes (just a feeling based off of vibes, don't take my word for it), so you need to constrain its scope more. That suits my workflow, half the time I have it generate code in the chat window and then write it myself, and I'm mostly using it at the level of generating function bodies and stuff, not entire features. Although it is writing a lot of SwiftUI without me really knowing the language and doing a fine job as far as I can tell (which isn't much admittedly).
One benefit I don't see talked about is it's speed - it's really quick, doesn't spend too much time reasoning even on "max", and the flash model is pretty dang good too. This lets me get into "flow state" when I'm writing code, compared to my experiences with Codex and Opus which would take minutes to complete even basic tasks and kind of ruined my focus.
It's so cheap though, you could download a different harness (Crush, OpenCode, Pi etc) and load $5 in credits and test it for yourself.
I'd recommend carefully looking at a few benchmarks (even though generally relying on benchmarks is problematic)
https://artificialanalysis.ai/evaluations/omniscience
Esp check the Hallucination rate for Deepseek - it's not good.
> Esp check the Hallucination rate for Deepseek - it's not good.
For strongly-typed coding tasks - and I imagine other tasks that have cheap validity checks: agentic harnesses and thinking tokens are an effective foil against hallucinations, at the expense of time. If a model hallucinates an API, compilation will fail and the error fed back into the machine so it can try again, in a two-steps-forward-one-step-back dance that is unreasonably effective. Given the price delta, it is often more cost effective to let the weaker model spiral towards a solution with many "Oh, wait..." turns
Yeah, the discounted deepseek inference is subsidized by the CCP for a reason, and it's one that might well come back to bite.
There is no evidence it is subsidized. Actually, there is evidence that (1) electricity is cheap in China & (2) deepseek is a very efficient model.
I think there is sufficient evidence to think its very likely. For example: https://www.americansecurityproject.org/wp-content/uploads/2...
> deepseek inference is subsidized by the CCP
What is that claim based on?
Besides common sense given the clear geopolitical context, sources like:
[1] https://chinaselectcommittee.house.gov/sites/evo-subsites/se... [2] https://ai.americansecurityproject.org/news/ai-imperative-20...
and more.
Of course, you can choose to ignore America-biased sources, but since it aligns with the obvious.
There is no evidence in those sources that DeepSeek is "subsidized" by the CCP in the way people imply (e.g. in an actively malicious*, market-distorting way that undercuts the competition, early Uber-style). They do receive tax breaks for their R&D research, a very common scheme in Europe (and which also used to be the case in the US, I believe). They also have public-private partnerships, e.g. the state is one of their clients. Also common in every free market economy. (SpaceX anyone?)
*This does not invalidate other concerns (censorship, privacy) but the way people phrase it makes it look like DeepSeek and co. are 'cheating' somehow with their business model by 'distorting' inference cost to make it way artificially lower than its 'natural price' (either notion being hopelessly naive)
"According to a report from Securities Times (a Chinese state-owned newspaper), Zhejiang Oriental, a listed company under the Zhejiang Provincial SASAC, participated in the angel round of financing of DeepSeek through its Hangzhou Oriental Jiafu Venture Capital Fund."[1]
"The Zhejiang Provincial State-owned Assets Supervision and Administration Commission (SASAC) is the provincial government agency in Zhejiang, China, responsible for managing, regulating, and overseeing the state-owned assets and enterprises owned by the provincial government." [2]
What does this imply? A state-owned company in China invested a ton of money into DeepSeek. aka State subsidization.
[1] https://www.americansecurityproject.org/wp-content/uploads/2... [2] https://www.fitchratings.com/research/corporate-finance/zhej...
They invested in a labelling company called "Deep Search" that news confused with "Deep Seek". It was corrected like a week later, of course very not agenda driven americansecuirtyproject never followed up / did retraction.
Have a source on that?
Too annoying to track down the original posts, but here's mirror:
>Gelonghui, February 11th | Zhejiang Orient Financial Holdings Group (600120.SH) announced the following explanation regarding the recently market-focused "DeepSeek Concept": DeepSeek is a large model under Hangzhou DeepSeek AI Basic Technology Research Co., Ltd. (hereinafter referred to as "DeepSeek"). In response to matters of concern in the Capital Markets, the company verified that as of the date of this announcement, the names of companies invested by the fund Sector managed by the company, such as Peking Deep Search Technology Co., Ltd. and Peking Jiuzhang Yunjike Technology Co., Ltd., are quite similar to those of DeepSeek and its affiliated enterprises, but there is no equity investment relationship. The company and the relevant private equity funds managed by the fund Sector have not directly or indirectly invested in DeepSeek.
ttps://news.futunn.com/en/post/53041547/zhejiang-orient-financial-holdings-group-600120-sh-and-its-managed?level=1&data_ticket=1780940972364876
Again, that's besides the point. So the state is an investor in DS, and? Many companies in Western capitalist economies receive initial state funding, especially startup grants. The real point to make is: does the state purposely fund the structural expenses of all those companies at a loss in an effort to undercut the competition and without which they would all go bankrupt and the cost of inference would be naturally much higher and couldn't be possibly optimized? I have yet to see evidence of that, especially given the continuous and prolific R&D from Chinese labs (or the panic at Meta when DS-r1 came out) that does show optimization gains are in fact possible.
An angel investor is an investor who provides early-stage capital to startups and entrepreneurs in exchange for ownership equity. That is not a grant or initial state funding. That is ownership. There are very few examples, especially prior to Trump, of government ownership/stakes of public companies.
But I will concede this: Due to the opaque nature of the Chinese economy to public scrutiny, we might never know.
I am sure, however that substantial use of Chinese inference (not their models per se, but on their servers) is, in aggregate, presents a substantial national security risk for the West. Heck, AI all by itself, without even considering other nations, is a national security threat of the near future, where national security is broadly construed as any threat against its people's welfare, no matter the actor.
>That is not a grant or initial state funding. That is ownership. There are very few examples, especially prior to Trump, of government ownership/stakes of public companies.
Maybe not in the US (although Musk getting state subsidies comes to mind), but very common in Europe. Quite a few founder friends of mine have gotten started with state funding (through various R&D promoting agencies). Angel investing is not the only startup funding structure out there
Check the pricing on OpenRouter. V4 Pro is twice as expensive from the next cheapest provider and 3.5x as expensive for fp8 (as opposed to fp4) from a US provider.
But I assume they're just harvesting training data since there's par for the course. There are also a handful of US labs offering free access for that exact reason.
Well, many people don't have very warm feelings for American LLM providers so they don't care. (Which matters because, at least anecdotally, they do care when buying a new car.)
also curious. On the claude code $200 plan, get close to weekly limits but don't usually hit it. to me just about any small reduction in performance would not be acceptable, the cost of redirecting and getting stuck during long runs without me are too big (like when I tried gemini cli for a few days).
if it's 99.9% comparable performance for less money I'm interested, but I'm skeptical it's there