I'm not sure you need a "DeepSeek native coding agent" to take advantage of DeepSeeks cache, yesterday as the Codex quota usage issue still wasn't solved for me, I wrote a tiny little bridge so I could use DeepSeek V4 Pro via Codex, and seems most of everything I did was basically cached as far as I can tell: https://i.imgur.com/7eKn6wN.png (2026-05-23 Input (Cache hit): 39,123,200 tokens, Input (Cache miss) 1,692,286), and the bridge is doing not special, just massage the DeepSeek API shape into what Codex expects, nothing particular about caching at all.
Besides being even better at the caching, I'm not sure what benefits you'd get compared to just firing up OpenCode with the DeepSeek API yourself, it'll similarly do caching for sure and also "talks directly to api.deepseek.com" if that matters, and you'll get a much more mature harness.
They explain some of the the reasons why they have a better solution and why they are very opinionated
>Automatic prefix caching activates only when the exact byte prefix of the previous request matches. Most agent loops reorder, rewrite, or inject fresh timestamps each turn — cache hit rate in practice: <20%.
So they optimize on this plus other techniques to improve cache hits, making it cheaper.
> tool call pruning breaks cache and people will tell you this is horrible and expensive
> except i looked at some anthropic data and real user behavior ends up with better cache hits and 30% less spend
> even this is needs to be analyzed further, it's just not simple
> for openai data it's inverted! cache hit ratio is actually better [sic: I think he meant worse based on the screenshot] with tool call pruning turned on
> but the net $ saved is only 5%
> kimi is a funny one - it has better cache hits with pruning on...but is also more expensive!
There was also another thread recently where he discussed that pruning improves user experience (models are smarter with less context) but I can't find it.
I can't confirm this. Having utilized Opencode for a large project over the past 10 months, with multiple models and agents, we've never run into such 'cache stability issues'."
> I wrote a tiny little bridge so I could use DeepSeek V4 Pro via Codex
Can you share the bridge. DeepSeek v4 is awesome paired with claude-code or opencode. I found that claude code costs me less than opencode and I am presuming this is due to a better engineered harness.
Sure, keep in mind it's a steaming pile of hacked together hacks, probably won't work in every case, doesn't support every feature that should be supported (like parallel tool calling, both Codex + DeepSeek API support it), and it might make your computer catch on fire: https://gist.github.com/embedding-shapes/eab3e63e5a95d3d78a2...
I only used it for a few hours to play around with stuff before the quota issue was fixed and I could resume using GPT models, and the bridge was coded by DeepSeek-V4-Flash-IQ2XXS + DwarfStar4 locally, I take no responsibility for what might happen with your computer or you, during usage or just reading the code.
Edit: heh, like don't look at line 117 for example where seemingly it likes to handle misspellings in the .env file which totally wasn't my fault for typo'ing the API key in that file... I'm sure there are tons of sharp edges and dumb stuff in there.
this appears to be native to the terminal, as in, there's no special application that runs or wraps an agent inside a tui. So basically instead of commands you type plain english?
> this appears to be native to the terminal, as in, there's no special application that runs or wraps an agent inside a tui
Same with codex? codex-rs at least, is a TUI as well, it does run a "app-server" in the background, that the TUI actually interacts with, but that's just an implementation detail. Also makes it easy to hook in your own programs to fire of codex "headless" sessions even without the TUI.
Not a fan of that page. The animated typing and resulting continuous resize of the example keeps moving the content beneath it down and up. Such bad UX.
Agents or no agents, people still need to test their websites on different resolutions or at least window width, but seems this is becoming a lost art.
High caching rates for coding agents can drastically reduce latency and API costs. I am curious to see how the caching strategy handles context invalidation across multiple files.
It's pretty funny, i'm a $200/m Claude subscriber and i've had little need to use anything else. However the more Claude has been restricting my workflow (notably around the recent IDE/-p usage change) the more i've been wanting to go elsehwere.
I'm concerned since i really want SOTA reasoning, but DeepSeek still has me interested.
> I'm concerned since i really want SOTA reasoning
I think you should give other models a try and see how much they differ from SOTA models. I did this and realized, even Qwen-2.5-Max was enough. I am sure even Claude Sonnet 3.5 is enough for things I play around with. I am not really striving for fields medal in Mathematics.
You should definitely stick to the $200 plan, and not try the $10 coding plans with open weight models and higher limits. Anthropic needs your money to stay solvent, and you'll sleep better knowing you're using SOTA.
> Hats off to the deekseek team for creating a great product
I have been using it for a while, and I wholeheartedly agree. imo, it is as good as codex or claude which I also use. It is a winner in the cost-sensitive tier, and if some startup could put it together with data-retention in mind, it could be a great product sold to the enterprise, as data-retention and privacy are the main issues for the coding-assistant usecase.
Deepseek v4 pro is definitely my preferred cheap model, it's very good, and I use it all the time for my personal projects (opencode go plan), but I also use Claude Opus all the time at work and Deepseek is not as good as that, but it does compete with Sonnet for capability, and beats it on price.
How can you have cache hit efficiency? Isn't it just a matter of not changing the previous context? I don't understand what knobs there are to tweak on this.
> Isn't it just a matter of not changing the previous context?
Yes, but a lot of harnesses change previous context. E.g. the system prompt injects the current time/date, working directory, files in the working directory, etc. Compaction also changes the whole previous context. I _think_ changing the list of tools also invalidates cache, so invoking a subagent with different tools would invalidate the cache.
My vague impression is that it's in a similar vein to functional programming languages. It generally disallows doing things that lead to bugs (cache misses in this case), and presumably allows you to do those things in a way that makes it much clearer that this is likely to cause cache misses. I would guess that in this paradigm, you don't mutate your existing session, you derive a new session by mutating the prior context into a new context.
>Can I point it at a self-hosted / private DeepSeek endpoint?
>Yes. Since 0.30 we accept non-standard key prefixes for self-hosted DeepSeek endpoints. Just point `baseUrl` at your internal address — the loop, cache strategy, and tool protocol are unchanged.
But my question is:
If I use Reasonix to talk to a deepseek endpoint through openrouter, am I still getting the cache-hit benifits of this agent harness?
Yes*. At least from my limited usage of deepseek-flash for a few billion tokens on openrouter, the cache-hit rate is >95%. And I simply used the claude code harness pointed at the openrouter anthropic compatible endpoint with no fluff.
> The loop is append-only, engineered around DeepSeek's byte-stable prefix cache — long sessions hold 90%+ cache hit and input-token cost collapses to ~1/5. Terminal-first, leave it running.
AI marketing slop. This is how all models and coding harnesses work, isn't it?
The author claims (in another AI-written post):
> LangChain — along with every generic agent framework I checked — rebuilds the prompt every turn. Timestamps get injected. History gets reordered. Tool schemas re-serialize with different whitespace.
I haven't touched LangChain in a long, long time, but don't think any of the current harnesses, Claude Code, Pi, Crush, OpenCode etc do that except if you change configuration? Keeping the context stable for caching is a very basic principle and not a wild innovation.
This posing as DeepSeek-specific is also a mystery.
Say you put the current time down to the second in the system prompt, which is the message that goes in front of the entire conversation, then basically nothing will be cached, every agent turn needs to ingest the entire session over and over. Contrast to not doing that, and the backend can leverage caching all the way up to the latest message, as nothing until then changed.
I don't think it's particularly effective to create a new coding agent when there's existing open-source agents (especially extremely extensible ones like Pi) that already optimize for cache hits, have far larger communities, and work for providers other than Deepseek.
I specifically use multiple different models and providers, so this wouldn't be useful for me.
And it contributes to the problem of each person vibe-coding their own, incompatible, half-baked tool in a space, instead of contributing to a small set of tools and expanding them.
Click on the download page, it's hilarious. It has a lot of information about the "smart probe" on the download and it's a realtime probe you can rerun.
That's the pinnacle of AI slop over engineered garbage in my opinion. All of that information is noise.
This is really tickling the conspiracy theorist part of my brain.
"Independent open-source project · not affiliated with DeepSeek"
"Reasonix only targets DeepSeek because..."
"Why DeepSeek only? Can I swap to Claude / GPT? It's a design choice, not a limitation"
The lady doth protest too much, methinks?
Nicely timed shortly after the making the rebate permanent anouncement.
Could just be Chinese devs trying to help western devs with some software and a western facing marketing campaign to raise awareness.
Could be DeepSeek astroturfing.
Could be "someone" in China trying to get more access to western data.
Is this really the behavior you want? Yes, doing tool-result clearing and such will blow your cache, but if you do it only occasionally, it's still likely a win. Yes, cache hits are good, but not so good that it's okay to be profligate with context to preserve those precious, precious KVs.
What AI model did you use for the website design? This is the second one I see with the exact same font and color scheme. Just curious because Claude models lean towards purples for example. Thank you!
This design still screams Claude to me, but a newer version than what you're thinking of. At some point they added a markdown file that tells it to use obviously AI designs like lots of blue/purple and gradients. Since then, this is its new style.
In my experience, it is claude-code paired with deepseek-v4. For penny-pinchers like me, I can have long coding sessions with it with no anxiety about the cost. Also, prompting it to what you want and verifying the outputs is more important than the quality of the model. So, I am better off with a cheaper model and taking the responsibility for prompting it and verifying the results.
I've gone through ~600m tokens in Xiaomi Mimo though Claude, and it's been the most effective use of an agent I've had yet. It's very capable, but generally not ambitious, picking simple but effective solutions to most problems I give it.
Going to write something longer about the experience when I get to a billion tokens.
Although I have little interest in agentic coding, when I do use it, I have found Kimi K2.6 to give Opus-quality output, and have switched entirely to it for pretty much everything.
I've used Opus extensively and tried K2.6 on a few projects, and the gap is huge. K2.6 is nowhere near the performance of Opus. That's fine because it's also far cheaper, but public benchmarks line up with my own personal experience that they aren't comparable in terms of intelligence.
(that is, different places on the Pareto efficiency graph)
I'd generally agree about Deepseek being as good as Sonnet - but I have extreme trouble with prompt compliance with V4 Pro in a way that I've never had with Sonnet. I'll tell it "find the bug, but don't fix it" or "please use this tool I just developed" and it'll ignore me a high fraction of the time.
It's bad enough that I'm working on guardrails at the harness level because prompting appears to be useless.
I'm not sure you need a "DeepSeek native coding agent" to take advantage of DeepSeeks cache, yesterday as the Codex quota usage issue still wasn't solved for me, I wrote a tiny little bridge so I could use DeepSeek V4 Pro via Codex, and seems most of everything I did was basically cached as far as I can tell: https://i.imgur.com/7eKn6wN.png (2026-05-23 Input (Cache hit): 39,123,200 tokens, Input (Cache miss) 1,692,286), and the bridge is doing not special, just massage the DeepSeek API shape into what Codex expects, nothing particular about caching at all.
Besides being even better at the caching, I'm not sure what benefits you'd get compared to just firing up OpenCode with the DeepSeek API yourself, it'll similarly do caching for sure and also "talks directly to api.deepseek.com" if that matters, and you'll get a much more mature harness.
This would be a better page to link to https://github.com/esengine/DeepSeek-Reasonix/blob/main/docs...
They explain some of the the reasons why they have a better solution and why they are very opinionated
>Automatic prefix caching activates only when the exact byte prefix of the previous request matches. Most agent loops reorder, rewrite, or inject fresh timestamps each turn — cache hit rate in practice: <20%.
So they optimize on this plus other techniques to improve cache hits, making it cheaper.
Opencode has really bad cache stability issues that they seem uninterested in fixing at the moment.
The OpenCode devs talk about this on Twitter a lot, e.g. https://xcancel.com/thdxr/status/2048268697790300343
> tool call pruning breaks cache and people will tell you this is horrible and expensive
> except i looked at some anthropic data and real user behavior ends up with better cache hits and 30% less spend
> even this is needs to be analyzed further, it's just not simple
> for openai data it's inverted! cache hit ratio is actually better [sic: I think he meant worse based on the screenshot] with tool call pruning turned on
> but the net $ saved is only 5%
> kimi is a funny one - it has better cache hits with pruning on...but is also more expensive!
There was also another thread recently where he discussed that pruning improves user experience (models are smarter with less context) but I can't find it.
This can also be disabled in the config: https://opencode.ai/docs/config/#compaction
I can't confirm this. Having utilized Opencode for a large project over the past 10 months, with multiple models and agents, we've never run into such 'cache stability issues'."
That'd be really easy to spot and also fix, most likely. Any open issue you could point us to, must surely been reported already?
5 replies →
[flagged]
> I wrote a tiny little bridge so I could use DeepSeek V4 Pro via Codex
Can you share the bridge. DeepSeek v4 is awesome paired with claude-code or opencode. I found that claude code costs me less than opencode and I am presuming this is due to a better engineered harness.
Sure, keep in mind it's a steaming pile of hacked together hacks, probably won't work in every case, doesn't support every feature that should be supported (like parallel tool calling, both Codex + DeepSeek API support it), and it might make your computer catch on fire: https://gist.github.com/embedding-shapes/eab3e63e5a95d3d78a2...
I only used it for a few hours to play around with stuff before the quota issue was fixed and I could resume using GPT models, and the bridge was coded by DeepSeek-V4-Flash-IQ2XXS + DwarfStar4 locally, I take no responsibility for what might happen with your computer or you, during usage or just reading the code.
Edit: heh, like don't look at line 117 for example where seemingly it likes to handle misspellings in the .env file which totally wasn't my fault for typo'ing the API key in that file... I'm sure there are tons of sharp edges and dumb stuff in there.
LiteLLM can serve OpenAI API endpoint IIRC and proxy that to other providers like DeepSeek, should work with Codex
I’m feeling more a novice every day, but how isn’t this just handing over your code to team deepseek for whatever they might want
4 replies →
this appears to be native to the terminal, as in, there's no special application that runs or wraps an agent inside a tui. So basically instead of commands you type plain english?
> this appears to be native to the terminal, as in, there's no special application that runs or wraps an agent inside a tui
Same with codex? codex-rs at least, is a TUI as well, it does run a "app-server" in the background, that the TUI actually interacts with, but that's just an implementation detail. Also makes it easy to hook in your own programs to fire of codex "headless" sessions even without the TUI.
Not a fan of that page. The animated typing and resulting continuous resize of the example keeps moving the content beneath it down and up. Such bad UX.
Agents or no agents, people still need to test their websites on different resolutions or at least window width, but seems this is becoming a lost art.
Yeah, doesn’t look designed for people who want to read it beyond animated typing animation.
Claude design AI slob.
High caching rates for coding agents can drastically reduce latency and API costs. I am curious to see how the caching strategy handles context invalidation across multiple files.
[flagged]
https://shittycodingagent.ai
Aka pi.dev
It's pretty funny, i'm a $200/m Claude subscriber and i've had little need to use anything else. However the more Claude has been restricting my workflow (notably around the recent IDE/-p usage change) the more i've been wanting to go elsehwere.
I'm concerned since i really want SOTA reasoning, but DeepSeek still has me interested.
> I'm concerned since i really want SOTA reasoning
I think you should give other models a try and see how much they differ from SOTA models. I did this and realized, even Qwen-2.5-Max was enough. I am sure even Claude Sonnet 3.5 is enough for things I play around with. I am not really striving for fields medal in Mathematics.
You should definitely stick to the $200 plan, and not try the $10 coding plans with open weight models and higher limits. Anthropic needs your money to stay solvent, and you'll sleep better knowing you're using SOTA.
If you want SOTA reasoning you should be using GPT 5.5 Pro.
Codex has only GPT 5.5
I love the focus on cache hit efficiency. Hats off to the deekseek team for creating a great product that maximizes cost efficiency for the user.
> Hats off to the deekseek team for creating a great product
I have been using it for a while, and I wholeheartedly agree. imo, it is as good as codex or claude which I also use. It is a winner in the cost-sensitive tier, and if some startup could put it together with data-retention in mind, it could be a great product sold to the enterprise, as data-retention and privacy are the main issues for the coding-assistant usecase.
Deepseek v4 pro is definitely my preferred cheap model, it's very good, and I use it all the time for my personal projects (opencode go plan), but I also use Claude Opus all the time at work and Deepseek is not as good as that, but it does compete with Sonnet for capability, and beats it on price.
Just in case, note that this project is someone's side project
> Independent open-source project · not affiliated with DeepSeek
Adding already cheap API cost and you probably could let it run for days and the same task..
How can you have cache hit efficiency? Isn't it just a matter of not changing the previous context? I don't understand what knobs there are to tweak on this.
> Isn't it just a matter of not changing the previous context?
Yes, but a lot of harnesses change previous context. E.g. the system prompt injects the current time/date, working directory, files in the working directory, etc. Compaction also changes the whole previous context. I _think_ changing the list of tools also invalidates cache, so invoking a subagent with different tools would invalidate the cache.
My vague impression is that it's in a similar vein to functional programming languages. It generally disallows doing things that lead to bugs (cache misses in this case), and presumably allows you to do those things in a way that makes it much clearer that this is likely to cause cache misses. I would guess that in this paradigm, you don't mutate your existing session, you derive a new session by mutating the prior context into a new context.
3 replies →
Okay, I'm curious.
From the FAQ, I see:
>Can I point it at a self-hosted / private DeepSeek endpoint?
>Yes. Since 0.30 we accept non-standard key prefixes for self-hosted DeepSeek endpoints. Just point `baseUrl` at your internal address — the loop, cache strategy, and tool protocol are unchanged.
But my question is: If I use Reasonix to talk to a deepseek endpoint through openrouter, am I still getting the cache-hit benifits of this agent harness?
Yes*. At least from my limited usage of deepseek-flash for a few billion tokens on openrouter, the cache-hit rate is >95%. And I simply used the claude code harness pointed at the openrouter anthropic compatible endpoint with no fluff.
thank you!
For those of you that use deepseek v4 occasionally, what harness do you use it with? I’m only familiar with claude code and codex.
Any comments on what you can or cannot rely on it for relative to cc and codex would be appreciated too!
Unusable thanks to the top animation pushing the rest of the site down repeatedly as you’re trying to read.
> The loop is append-only, engineered around DeepSeek's byte-stable prefix cache — long sessions hold 90%+ cache hit and input-token cost collapses to ~1/5. Terminal-first, leave it running.
AI marketing slop. This is how all models and coding harnesses work, isn't it?
The author claims (in another AI-written post):
> LangChain — along with every generic agent framework I checked — rebuilds the prompt every turn. Timestamps get injected. History gets reordered. Tool schemas re-serialize with different whitespace.
I haven't touched LangChain in a long, long time, but don't think any of the current harnesses, Claude Code, Pi, Crush, OpenCode etc do that except if you change configuration? Keeping the context stable for caching is a very basic principle and not a wild innovation.
This posing as DeepSeek-specific is also a mystery.
Wow the UI looks exactly what I vibe coded yesterday. What a coincidence
It's obvious why...
I would've liked benchmarks against other harnesses showing the caching performance
Is there benchmarks and measurements that offers comparisons between different harnesses?
Good timing given the cost spike across other frontier models.
Good thing DS just made their discount permanent. https://x.com/deepseek_ai/status/2057854261699195173
Can someone give me a eli5 version of what this is? It really sounds useful to Claude subscribers.
Is this improving the cache hit and hence overall efficiency of coding workflows?
Does it also let me host a local llm (deepseek)? What are model min requirements for this?
You can also ask Claude and get an immediate answer, the power is yours
Isn't caching a server-side thing? How does the agent affect it, significantly at least?
Say you put the current time down to the second in the system prompt, which is the message that goes in front of the entire conversation, then basically nothing will be cached, every agent turn needs to ingest the entire session over and over. Contrast to not doing that, and the backend can leverage caching all the way up to the latest message, as nothing until then changed.
Surely other agent CLIs are not dumb enough to invalidate cache on every turn over something so obvious?
3 replies →
Yes, of course you can destroy it. But how far can you "improve", beyond decent "common sense" behaviour.
I don't think it's particularly effective to create a new coding agent when there's existing open-source agents (especially extremely extensible ones like Pi) that already optimize for cache hits, have far larger communities, and work for providers other than Deepseek.
I specifically use multiple different models and providers, so this wouldn't be useful for me.
And it contributes to the problem of each person vibe-coding their own, incompatible, half-baked tool in a space, instead of contributing to a small set of tools and expanding them.
It'd be better to just extend an existing tool.
Click on the download page, it's hilarious. It has a lot of information about the "smart probe" on the download and it's a realtime probe you can rerun.
That's the pinnacle of AI slop over engineered garbage in my opinion. All of that information is noise.
Doesn't Pi Agent do exactly this? Assuming "append only" means they do some kind of compaction as well.
This is really tickling the conspiracy theorist part of my brain.
"Independent open-source project · not affiliated with DeepSeek" "Reasonix only targets DeepSeek because..." "Why DeepSeek only? Can I swap to Claude / GPT? It's a design choice, not a limitation"
The lady doth protest too much, methinks?
Nicely timed shortly after the making the rebate permanent anouncement.
Could just be Chinese devs trying to help western devs with some software and a western facing marketing campaign to raise awareness. Could be DeepSeek astroturfing. Could be "someone" in China trying to get more access to western data.
Who knows?
But Claude made the website?
What conclusion are you drawing from that?
> no reordering, no marker-based compaction
Is this really the behavior you want? Yes, doing tool-result clearing and such will blow your cache, but if you do it only occasionally, it's still likely a win. Yes, cache hits are good, but not so good that it's okay to be profligate with context to preserve those precious, precious KVs.
you created it yourself ?
No.
This Claude front end skill is now soon to be slop.
Oh, I was wondering why all new websites look shitty in the same way.
Already is. Every new website looks exactly the same.
What AI model did you use for the website design? This is the second one I see with the exact same font and color scheme. Just curious because Claude models lean towards purples for example. Thank you!
Opus 4.7 selects such palette and motifs by default. Might even be first iteration of claude design.
This design still screams Claude to me, but a newer version than what you're thinking of. At some point they added a markdown file that tells it to use obviously AI designs like lots of blue/purple and gradients. Since then, this is its new style.
DeepSeek v4 perhaps?
Frontend design skill by Anthropic specifically says not to use purple. I'd be surprised if it still uses purple. Have you seen that recently?
So what's best low cost coding agent these days? Kimi 2.6? Qwen's latest closed model? Composer 2.5? DeepSeek?
In my experience, it is claude-code paired with deepseek-v4. For penny-pinchers like me, I can have long coding sessions with it with no anxiety about the cost. Also, prompting it to what you want and verifying the outputs is more important than the quality of the model. So, I am better off with a cheaper model and taking the responsibility for prompting it and verifying the results.
It's obviously much cheaper paying by the token but how does it compare to a codex subscription on cost?
Can you quantify the actual costs in a week and the use you make?
1 reply →
I've gone through ~600m tokens in Xiaomi Mimo though Claude, and it's been the most effective use of an agent I've had yet. It's very capable, but generally not ambitious, picking simple but effective solutions to most problems I give it. Going to write something longer about the experience when I get to a billion tokens.
I do have my eyes on the coding plan, which is quite generous.
https://mimo.mi.com
Are you using Mimo 2.5 pro?
1 reply →
Although I have little interest in agentic coding, when I do use it, I have found Kimi K2.6 to give Opus-quality output, and have switched entirely to it for pretty much everything.
I've used Opus extensively and tried K2.6 on a few projects, and the gap is huge. K2.6 is nowhere near the performance of Opus. That's fine because it's also far cheaper, but public benchmarks line up with my own personal experience that they aren't comparable in terms of intelligence.
(that is, different places on the Pareto efficiency graph)
Seems to be DeepSeek.
https://news.ycombinator.com/item?id=48237663
Kimi 2.6 is great. Qwen3.7-max benchmarks similarly but I havent used it yet
For me, it's by far Deepseek. It's many times cheaper than competitors, and about as good as Sonnet 4.6.
I'd generally agree about Deepseek being as good as Sonnet - but I have extreme trouble with prompt compliance with V4 Pro in a way that I've never had with Sonnet. I'll tell it "find the bug, but don't fix it" or "please use this tool I just developed" and it'll ignore me a high fraction of the time.
It's bad enough that I'm working on guardrails at the harness level because prompting appears to be useless.
Do you have the same issue?
1 reply →
Just use codex with 5.5 on low reasoning levels
[dead]
[dead]