Codex CLI's Bedrock provider bug inflates GPT-5.6 costs 10x
A GitHub issue shows OpenAI Codex can't enable explicit prompt caching on AWS Bedrock, driving huge cache-write bills for agentic coding sessions.
What to know
- Codex CLI's native Amazon Bedrock provider cannot configure explicit prompt caching for GPT-5.6 Sol, forcing costly full-prefix cache rewrites on agentic coding workloads.
- Real production usage over four days showed cache-write tokens made up ~85% of estimated Bedrock spend, roughly matching the reported 10x cost increase.
- The report requests Codex add prompt_cache_options/prompt_cache_breakpoint support and expose per-turn cache read/write telemetry so users can diagnose costs.
- The issue reached Hacker News's front page (136 points, 51 comments) and was reposted on Mastodon, but no fix or maintainer response is documented in the evidence.
“This report does not claim that every cache write is a defect.”
GitHub issue #37674 author, Codex user reporting the bug · GitHub · Aug 19
TheP1000 GitHub issue reporter / HN submitterOpenAI Maintainer of Codex CLIAmazon Web Services (Bedrock) Cloud host of the GPT-5.6 Sol model
The record 1 articles and posts · last 30 days
What people are saying 12 voices from 1 site · best of 13 · verbatim
- Aug 21
-
You know, installing unsloth studio lets you use codex against a local qwen 3.8 instance, which does ~10 tok/sec without GPU on a modern machine, and 100+ tok/sec on a 5090, and is incredibly good.
-
I love the fact that devs are still complaining that invoices are able to grow from $300 to $1000+How can anyone use a platform where this is even an issue? Just because AWS is a failure in this regard, doesn't mean there aren't alternatives with fixed prices or others with easily settable limits.
-
> Wow, that whole thread is borderline incoherent, presumably generated by an AI without adequate oversight.What sucks is that every issue tracker for these agent harnesses are the same, and this shit hides real issues!For example, Codex started encrypting messages from a agent to the sub-agents when you use Sol + Ultra, which is terrible for…
-
But but according to Mr Altman, we've like entered the singularity. Right? Who cares about a billing issue?
-
Prompt edits leaking into the cache and affecting model responses is exactly the kind of billing-relevant behavior change that should be in release notes, not discovered by users.
-
Codex usage feels exorbitantly high since today. They [0] are denying it, but the number of anecdotal users who decided to raise this as an issue (as a result it's trending on X) says otherwise.[0]
-
Rookie mistake - it seems like they didn't follow manufacturers' guidance when installing the 10x engineers. One needs to clearly define which metric should be 10x'd before powering them up.
-
I have noticed the same thing starting with 5.6 when editing my last prompt inside the vscode codex plugin, I’ve seen the model’s thinking respond to the edit with a remark.Slightly bummed out about it because in the past you could try different situations during a planning session and it wouldn’t pollute the cache but now it does. I’m not sure if…
-
The fact that companies with access to SOTA non-public AI keep having these kinds of dumb bugs in something as simple as a chat app helps validate my disbelief of people claiming AI is ready to replace all software development.
-
> It sounds to be like 5.5 and below may have been using a conventional attention scheme where a cached KV sequence can be easily used to restore a prefix of itself, but perhaps 5.6 is using linear attention or LSTM or another recurrent scheme where you cannot rewind the model state by just truncating it.I feel like this is the kind of substantial…
-
It’s really cool that we have this proof that US companies are half year behind Chinese models in architecture.
- Aug 20
-
Our codex on AWS Bedrock read / write cache ratio was less than 5%. Cache writes are very expensive and they were never being used. This results in codex on Bedrock causing ~10x what it should due to no caching and massive writes.The workaround in issue resolved for me: web_search = "disabled"