I do my own framework and spend a lot of time trying to debug this and it’s not so much the context size in hard numbers but rather the probability that there is debris or wrong directions in the window that are drowning out the things the user thinks are important.
This manifests in the llm that keeps going back to doing the thing that failed when they tried it just before the last approach etc. The frequency of things in the context window give weight even if they are the wrong things.
I have a lot of tricks like not giving the llm lots of tools but rather giving it a tool it can use to search for tools etc.
But the bigger solution is in process where you use something like superpowers to force the llm through stages and you control the context that carries forward.
I've been able to avoid context size issues by applying one simple constraint to my agent loop. What I do is prevent all tool calling in the user's top-level conversation thread. Anything that needs to tool call must happen in a recursive invoke of the agent, which returns whatever results to caller.
I can keep the same high level conversation going for an entire day over a million LOC+ codebase without ever hitting meaningful token limits. No compaction or summarization tricks needed. I can burn 50 million tokens in recursive calls and still not touch 100k tokens in my root conversation thread.
There is some rework needed to "bootstrap" the agent each time it has to descend back into Narnia, but this is still far more efficient than carrying around one big flat context that tries to cover everything all the time.
Recursion is very effective at controlling token use, but it can only go so far. I've not observed any uplift for recursive depth beyond 1. I have seen the agent attempt it a few times, but the practical performance is simply not there. External symbolic recursion does not appear to be something the frontier models have been trained for. They are fantastic at emulating recursion in context, but we don't want that if we are trying to achieve a reduction in token use.
This has not been my experience with Opus since Anthropic released the 1M token context window for use under the subscription plans. I routinely push past 500k tokens, even sometimes up to around 800k tokens, and don't see this problem. I've seen it to some extent when getting truly near the limit, up around and above 900k tokens, though what I see isn't as severe as the author seems to see.
(And I rarely fill the context window that far anyway when working on a single task, or a series of tasks that are related enough to warrant the same context; more typical is anywhere between 200k and 600k or so.)
I'm not saying that no one ever has this experience, but it's odd to me that some people see it so often that it warrants giving it a name.
I'm always a bit confused when people say things like this. 60k token is often more than the initial context I feed the model with. And I don't think I ever had a productive session that began under 150k tokens.
I hate to do the "you're holding it wrong" trope, but I think you might have something misconfigured somewhere unless you missed a 0, because just past 60k tokens is such a small context window to be seeing issue in.
Do you have any old documentation that it's picking up and referencing? If you set all claude settings back to default do you see the same issue?
Opus 4.6 was on drugs past 200k, I skipped 4.7, 4.8 did good up to ~350k, and Fable did great beyond 400k, in my limited testing. The quality does appear to be trending upwards.
I have a custom build command for a rust project (yarn build:lib) and my experience is 120k for GLM and roughly 200-300k for Opus. After that, they default to cargo build.
My projects have specific build/verify steps as well, and after a certain point Claude forgets to run them. I’m going to try a “No brown M&Ms” hook to halt Claude if it tries to run the default command instead of the instructed commands from CLAUDE.md. Perhaps this will be a good signal that a compacted or fresh session is needed at that point to avoid mistakes.
I'm getting a lot of mileage out of basically acting like the AI's Product Manager, and insisting that it writes up short PRDs for every feature we propose to build. That gives it a reference over time of everything that has been built, but also makes it less liable to drift with each one. Each one gets its own conversation. For me this is a happy medium between stopping it going off the rails but also making sure it can reference past decisions when it needs to. The one thing I dislike about Pocock's method (not to use PRDs so much but to have an in depth discussion to get alignment) first is it wastes a lot of the best window on that initial back and forth.
Is it adhoc or you use more structured approaches like openspec? I also tend to work on a plan first, but it stays as in-session todo, which is hard to reference later.
It's ad hoc / my own framework, just found something which works for me. The exact structure is
- Work Mode - HITL/AFK
- Problem Statement
- Who It Affects - Primary / Secondary User
- User Stories
- Business Case
- Why Now
- Success Critera
- In Scope/Out of Scope [Out of Scope v. important)
- Thinnest Slice (This I've found super valuable, means you max out the amount of 'product' for your buck and avoid diminishing marginal returns or overbuilding. Often I will build this)
- Eigenfeature - What is the larger feature we _could_ (but probably won't) which would solve for this use case and other stuff I might not have thought of
- Technical Notes
- Deps
- Schema Changes
- Risks
- Final Recommendation [go / no go, including on scope]
There's a note in my Claude / Agents MD which says no net new feature gets introduced without this and I get it to move through a pipeline of folders (active, approved, shipped, proposed etc). All runs in a system of MD files and have even created a little MD Kanban from the metadata!
I guess I've stumbled into something similar. Though I don't have a fixed format like yours. I first do a lot of back and forth to generate what I call a design document also includes rationales for various points or decisions. I use both Claude and Codex to iterate on this until I'm happy. The end result includes a lot of what you mention.
I then start a fresh conversation, make it analyze the design document and code, and for larger changes, generate a high-level implementation document which includes concrete phases or steps. I review this plan and iterate if necessary.
Then for each phase I make it generate a detailed plan for that phase and save it along side the other documents. Once the phase is over, I make it write a summary of what was done, decisions made and reasons for it. And typically a good point to compact the model's context.
These documents gives additional context for when I make another model do code review, and help illuminate drift or gaps from the main design document.
I wonder how much this depends on the quality and consistency of the context?
For example, it may be the case that a long context full of useful information relevant to the task is completely fine, perhaps even beneficial. And if the context contains a bunch of unrelated tangents and conflicting instructions, then it will be detrimental.
Have there been studies on what makes models get dumber? To what extent is context length to blame vs context quality?
Opus in recent versions is fine beyond 100k, but I usually do try to keep it under 200k.
But, this is also why so-called "memory" systems are usually a mistake that make the models dumber. They don't have memory, they only have context, and every irrelevant fact you shove into the context is less context for the problem. Less distractions, better results.
The way to have the agent remember things is to have it document its work, like a human developer would do if they wanted their project to be friendly to other developers working on it. Good developer docs with an index page and a good plan with checklists, in concise Markdown files, checked in to the repo is the ideal memory for models and the ideal docs you need to figure out WTF the model has been up to. Helps with code review, too, whether by humans or another model. There's no down side.
At least for me, Opus keeps writing stuff to memories, only to consistently forget checking those memories before doing the same mistake again. This ("remember to check memories!") is of course then again written as a memory... Clearly not a very well working system, yep.
Considerations about what goes on in agents internally will probably not be part of software development for long.
Personally, I already see LLMs and agents as blackboxes. I give each feature request to multiple LLMs and then compare the results. I don't manually use "sessions" at all. I just look at the outcome. When I dislike it, I "git reset --hard", change my prompts and restart the feature request.
To have an ongoing sense of which agents perform best, I keep a log and calculate an ELO score of which agents meet my demands best. This score is imporant to me, not so much how the agent achieves it.
Unless we do our own benchmarks, we have to take all the marketing fluff from the frontier labs at face value, and all public benchmarks degrade eventually as labs optimize towards them. OP’s approach is wasteful because it is brute force, but post says that an ELO is kept, so this is also an experiment, and I don‘t see what‘s wrong with that. You learn which model performs well in which settings which may save resources later. It‘s also wasteful to keep working with the wrong model/harness/tools for too long.
In an interactive session, adding "Fine, but make the button red" after the model generated a first solution more than doubles the tokens used. As the model now not only gets the original code and the feature request but also the updated code plus the change request as input tokens.
Sending a feature request to an LLM and then sending the feature request again with "The button shall be red" only doubles the tokens used.
I wrote my own agent, and it sends data to LLMs in this order: "General Prompts (How to write good code)" + "The Code" + "The Feature Request". This means the KV cache will be used even when the feature request changes.
And output tokens are usually way less than the input tokens.
So I think that my approach is very lightweight on token usage compared to an interactive session.
It would be interesting to measure it for the other agents out there. Sending a feature request two times vs an interactive session.
That’s usually not true due to caching. It may be true if you leave a large gap in between, but if you send “make it red” right after, then it’s purely incremental
I've had no problem with Claude Code Opus 4.8 effort max using 20% token context (200k) on software development tasks (all stages). I aways load core source files and the ones we are working on up front. Around 20%, I make it autoprepare for a new session and clear.
Admittedly I have been doing this precautiously, based on anecdotal evidence, not because I had bad experiences with longer context deterioration myself.
In the brief time I had access to Fable 5, it went on long running tasks (>45 mins) into the 30-40% zone without apparent context coherence problems.
The approach we're taking to deal with this very real context rot is using a bunch of related techniques which we call transposing the agent loop: https://alejo.ch/3jt
In essence, we run many short agent loops, generating their prompts dynamically from structured data. Each loop advances the state in a small step towards the final goal.
In my own testing I have seen peak performance happen usually within 15-20% of the intended context limit, albeit there are a few optimizations depending on the task quality.
I /clear all the time out of habit. I want to be able to get the thing done with minimal context. It also means you can do it again slightly different if needed, you know the seed conditions for the task.
i let the main loop spawn sub terminal via tmux to prevent large contexts. it's great to divide tasks in small patterns and consolidate it step by step.
There's an env var you can set in Claude Code to bring the autocompact threshold down, effectively setting your own max context window. I have it at 400k.
Even taking the author's criticism about large context windows for granted, which in my experience are exaggerated, they are still a huge UX improvement over short windows. That reason alone is enough for me to support them.
Perhaps compacting the context can be made in multiple requests over smaller and overlapping chunks to avoid using the 'dumb zone', and for yielding a better result.
Yes context management is key.
I do my own framework and spend a lot of time trying to debug this and it’s not so much the context size in hard numbers but rather the probability that there is debris or wrong directions in the window that are drowning out the things the user thinks are important.
This manifests in the llm that keeps going back to doing the thing that failed when they tried it just before the last approach etc. The frequency of things in the context window give weight even if they are the wrong things.
I have a lot of tricks like not giving the llm lots of tools but rather giving it a tool it can use to search for tools etc.
But the bigger solution is in process where you use something like superpowers to force the llm through stages and you control the context that carries forward.
I've been able to avoid context size issues by applying one simple constraint to my agent loop. What I do is prevent all tool calling in the user's top-level conversation thread. Anything that needs to tool call must happen in a recursive invoke of the agent, which returns whatever results to caller.
I can keep the same high level conversation going for an entire day over a million LOC+ codebase without ever hitting meaningful token limits. No compaction or summarization tricks needed. I can burn 50 million tokens in recursive calls and still not touch 100k tokens in my root conversation thread.
There is some rework needed to "bootstrap" the agent each time it has to descend back into Narnia, but this is still far more efficient than carrying around one big flat context that tries to cover everything all the time.
Recursion is very effective at controlling token use, but it can only go so far. I've not observed any uplift for recursive depth beyond 1. I have seen the agent attempt it a few times, but the practical performance is simply not there. External symbolic recursion does not appear to be something the frontier models have been trained for. They are fantastic at emulating recursion in context, but we don't want that if we are trying to achieve a reduction in token use.
This has not been my experience with Opus since Anthropic released the 1M token context window for use under the subscription plans. I routinely push past 500k tokens, even sometimes up to around 800k tokens, and don't see this problem. I've seen it to some extent when getting truly near the limit, up around and above 900k tokens, though what I see isn't as severe as the author seems to see.
(And I rarely fill the context window that far anyway when working on a single task, or a series of tasks that are related enough to warrant the same context; more typical is anywhere between 200k and 600k or so.)
I'm not saying that no one ever has this experience, but it's odd to me that some people see it so often that it warrants giving it a name.
I see this said often and find it insane given how many times I find opus models making basic recall mistakes at <100k tokens.
Personally I consider < 60k to be the smart zone for opus. This is worse for opus 4.7 and 4.8 cause of the more granular tokenizer
I'm always a bit confused when people say things like this. 60k token is often more than the initial context I feed the model with. And I don't think I ever had a productive session that began under 150k tokens.
60k is tiny, if it's making recall mistakes that early then you might have some false memories or incorrect instructions in your CLAUDE.md.
60k isn't much bigger than the system prompt.
Yeah 60k is ludicrous, I've barely seeded the context at that point and I don't see context related degradation until well into the 600-700k.
>you might have some false memories or incorrect instructions in your CLAUDE.md
did you internalize what was wrong with that quote when it was said? does it apply here?
[dead]
Not specific to Opus but yes it would make mistakes. I usually try to keep context window under 10%
I hate to do the "you're holding it wrong" trope, but I think you might have something misconfigured somewhere unless you missed a 0, because just past 60k tokens is such a small context window to be seeing issue in.
Do you have any old documentation that it's picking up and referencing? If you set all claude settings back to default do you see the same issue?
Opus 4.6 was on drugs past 200k, I skipped 4.7, 4.8 did good up to ~350k, and Fable did great beyond 400k, in my limited testing. The quality does appear to be trending upwards.
I’ve had similar experiences with Fable. 70%+ context used out of 1M, still sharp and no memory issues.
Thats another problem of this post, the author mentions Claude but not explicitely what models...
100k tokens "by lunch" is also not my finding, the newer models will hit that already right in the initial exploratory phase
Really depends on the project.
I found "by lunch" odd too, but considering that Claude wrote the article, it's not going to know specifics.
I have a custom build command for a rust project (yarn build:lib) and my experience is 120k for GLM and roughly 200-300k for Opus. After that, they default to cargo build.
My projects have specific build/verify steps as well, and after a certain point Claude forgets to run them. I’m going to try a “No brown M&Ms” hook to halt Claude if it tries to run the default command instead of the instructed commands from CLAUDE.md. Perhaps this will be a good signal that a compacted or fresh session is needed at that point to avoid mistakes.
As the gamblers say at the poker table: If you can't figure out who the mark is when you site down...
Evaluating the Sensitivity of LLMs to Prior Context
https://arxiv.org/abs/2506.00069
I'm getting a lot of mileage out of basically acting like the AI's Product Manager, and insisting that it writes up short PRDs for every feature we propose to build. That gives it a reference over time of everything that has been built, but also makes it less liable to drift with each one. Each one gets its own conversation. For me this is a happy medium between stopping it going off the rails but also making sure it can reference past decisions when it needs to. The one thing I dislike about Pocock's method (not to use PRDs so much but to have an in depth discussion to get alignment) first is it wastes a lot of the best window on that initial back and forth.
Is it adhoc or you use more structured approaches like openspec? I also tend to work on a plan first, but it stays as in-session todo, which is hard to reference later.
It's ad hoc / my own framework, just found something which works for me. The exact structure is
- Work Mode - HITL/AFK
- Problem Statement
- Who It Affects - Primary / Secondary User
- User Stories
- Business Case
- Why Now
- Success Critera
- In Scope/Out of Scope [Out of Scope v. important)
- Thinnest Slice (This I've found super valuable, means you max out the amount of 'product' for your buck and avoid diminishing marginal returns or overbuilding. Often I will build this)
- Eigenfeature - What is the larger feature we _could_ (but probably won't) which would solve for this use case and other stuff I might not have thought of
- Technical Notes
- Deps
- Schema Changes
- Risks
- Final Recommendation [go / no go, including on scope]
There's a note in my Claude / Agents MD which says no net new feature gets introduced without this and I get it to move through a pipeline of folders (active, approved, shipped, proposed etc). All runs in a system of MD files and have even created a little MD Kanban from the metadata!
I guess I've stumbled into something similar. Though I don't have a fixed format like yours. I first do a lot of back and forth to generate what I call a design document also includes rationales for various points or decisions. I use both Claude and Codex to iterate on this until I'm happy. The end result includes a lot of what you mention.
I then start a fresh conversation, make it analyze the design document and code, and for larger changes, generate a high-level implementation document which includes concrete phases or steps. I review this plan and iterate if necessary.
Then for each phase I make it generate a detailed plan for that phase and save it along side the other documents. Once the phase is over, I make it write a summary of what was done, decisions made and reasons for it. And typically a good point to compact the model's context.
These documents gives additional context for when I make another model do code review, and help illuminate drift or gaps from the main design document.
Is there back-and-forth? How long do these get? Can you share an example?
I wonder how much this depends on the quality and consistency of the context?
For example, it may be the case that a long context full of useful information relevant to the task is completely fine, perhaps even beneficial. And if the context contains a bunch of unrelated tangents and conflicting instructions, then it will be detrimental.
Have there been studies on what makes models get dumber? To what extent is context length to blame vs context quality?
Opus in recent versions is fine beyond 100k, but I usually do try to keep it under 200k.
But, this is also why so-called "memory" systems are usually a mistake that make the models dumber. They don't have memory, they only have context, and every irrelevant fact you shove into the context is less context for the problem. Less distractions, better results.
The way to have the agent remember things is to have it document its work, like a human developer would do if they wanted their project to be friendly to other developers working on it. Good developer docs with an index page and a good plan with checklists, in concise Markdown files, checked in to the repo is the ideal memory for models and the ideal docs you need to figure out WTF the model has been up to. Helps with code review, too, whether by humans or another model. There's no down side.
At least for me, Opus keeps writing stuff to memories, only to consistently forget checking those memories before doing the same mistake again. This ("remember to check memories!") is of course then again written as a memory... Clearly not a very well working system, yep.
Considerations about what goes on in agents internally will probably not be part of software development for long.
Personally, I already see LLMs and agents as blackboxes. I give each feature request to multiple LLMs and then compare the results. I don't manually use "sessions" at all. I just look at the outcome. When I dislike it, I "git reset --hard", change my prompts and restart the feature request.
To have an ongoing sense of which agents perform best, I keep a log and calculate an ELO score of which agents meet my demands best. This score is imporant to me, not so much how the agent achieves it.
This is an absolutely crazy wasteful thing to do considering the actual cost of all that inference and nothing to be proud of.
Unless we do our own benchmarks, we have to take all the marketing fluff from the frontier labs at face value, and all public benchmarks degrade eventually as labs optimize towards them. OP’s approach is wasteful because it is brute force, but post says that an ELO is kept, so this is also an experiment, and I don‘t see what‘s wrong with that. You learn which model performs well in which settings which may save resources later. It‘s also wasteful to keep working with the wrong model/harness/tools for too long.
It is the other way round.
In an interactive session, adding "Fine, but make the button red" after the model generated a first solution more than doubles the tokens used. As the model now not only gets the original code and the feature request but also the updated code plus the change request as input tokens.
Sending a feature request to an LLM and then sending the feature request again with "The button shall be red" only doubles the tokens used.
The cost is far from linear though. Because of prompt caching and the fact that generally output tokens are a lot more expensive than input tokens.
Agreed that it is not linear.
I wrote my own agent, and it sends data to LLMs in this order: "General Prompts (How to write good code)" + "The Code" + "The Feature Request". This means the KV cache will be used even when the feature request changes.
And output tokens are usually way less than the input tokens.
So I think that my approach is very lightweight on token usage compared to an interactive session.
It would be interesting to measure it for the other agents out there. Sending a feature request two times vs an interactive session.
"Make the button red" probably doesn't need an LLM at all.
That’s usually not true due to caching. It may be true if you leave a large gap in between, but if you send “make it red” right after, then it’s purely incremental
Probably like 1% of the energy an average person spends on driving.
Average american is what you mean
come on now, we can't just not escape the permanent underclass by using our brains, we've also got to use up all the resources while doing it.
[delayed]
I've had no problem with Claude Code Opus 4.8 effort max using 20% token context (200k) on software development tasks (all stages). I aways load core source files and the ones we are working on up front. Around 20%, I make it autoprepare for a new session and clear.
Admittedly I have been doing this precautiously, based on anecdotal evidence, not because I had bad experiences with longer context deterioration myself.
In the brief time I had access to Fable 5, it went on long running tasks (>45 mins) into the 30-40% zone without apparent context coherence problems.
The approach we're taking to deal with this very real context rot is using a bunch of related techniques which we call transposing the agent loop: https://alejo.ch/3jt
In essence, we run many short agent loops, generating their prompts dynamically from structured data. Each loop advances the state in a small step towards the final goal.
In my own testing I have seen peak performance happen usually within 15-20% of the intended context limit, albeit there are a few optimizations depending on the task quality.
I /clear all the time out of habit. I want to be able to get the thing done with minimal context. It also means you can do it again slightly different if needed, you know the seed conditions for the task.
i let the main loop spawn sub terminal via tmux to prevent large contexts. it's great to divide tasks in small patterns and consolidate it step by step.
There's an env var you can set in Claude Code to bring the autocompact threshold down, effectively setting your own max context window. I have it at 400k.
Even taking the author's criticism about large context windows for granted, which in my experience are exaggerated, they are still a huge UX improvement over short windows. That reason alone is enough for me to support them.
Perhaps compacting the context can be made in multiple requests over smaller and overlapping chunks to avoid using the 'dumb zone', and for yielding a better result.
Even better, don't trust LLMs at all.
Is there any chance that this is because training corpus largely consists of documents shorter than the advertised context windows?
Hasn’t been my experience at all - 1M window is a very clear upgrade working with Claude code.
[flagged]
[flagged]
[flagged]