Rendered at 22:32:32 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
xlayn 2 hours ago [-]
If you use Qwen3.8 27B to manage, to orchestrate, do do shallow research, and then make it ask claude you will notice that Qwen can get out of Haiku, Sonnet and Opus almost as good answers... I was researching how good the model asnswers where, I can't get to work with anything but Fable, but Qwen can get good work from any of them.
My theory is that if you make Qwen understand you, Qwen can elaborate as much as needed to get any of those models get good results, because it's filling in details.
So using Qwen to direct claude also get's you amazing results way way cheaper...
Update: let me elaborate more, I read the HN entry about the team of 5 Qwens being close to fable quality... so I embarked to add to my harness (https://github.com/alainnothere/privibe because we all have one right?) and once I had added the orchestrate tool I had to now measure somehow it it's able to improve quality, so I claude and then Qwen to run a experiment to call my harness passing the model to use and ask them to do something, 3 rounds I think and see what quality came out of it... and as per the grading... qwen and claude agreed that the quality of the response comming out the harness using the orchestrate was VERY CLOSE, with opus and fable providing more "in my opinion you should" which matches my experience on why they are better.
One more update: I can't get copilot models to work decently, it may be my way of working... but I can get Qwen to follow what I try to do... if Qwen get's me he can then fire copilot cli and get amazing results by using copilot... bonus points seems like copilot cli is cheaper with the AIC (cost of the requests) so you can make the copilot subscription longer
drkleiner 2 hours ago [-]
I do the opposite of this, use Qwen3.8-27B as subagents, Opus 5.5 as orchestrator and planner, never implementer, and it cuts my token usage significantly. So it is using Claude to direct Qwen. Never thought of the other way around, will give this a try soon as possible.
ihuman 2 hours ago [-]
What do you use to get Opus to call Qwen?
drkleiner 1 hours ago [-]
Custom MCP with a variety of functions, I have different Qwen profiles with a variety of subagent numbers and cache pools, my default is 192k pool with 64k initial context that can grow 32k at a time up to 128k, allowing 3x64k, 2x96k or 1x128k and 1x64k subagents. Qwen uses Pi harness. Model I use specifically is ukisai/Swift-1.5-Qwen3.8-27B-GSQ-RCO-GGUF
nullstyle 1 hours ago [-]
ohmypi will spawn subagents across different providers out of the box. Opus will spawn subagents running on my local ds4 installation just fine
verdverm 2 hours ago [-]
You can have a setup like this with OpenCode agents, where you can define custom top-level agents, subagents, and assign specific models to each.
There are a lot of ways to do this if you stop using Big Ai harnesses that make it difficult to use models besides theirs.
irthomasthomas 2 hours ago [-]
Install the llm cli and tell opus to use it to prompt qwen.
rao-v 51 minutes ago [-]
Oh man I’m been playing with little -> big vs big -> little sub-tasking a bunch.
My sense so far is it’s very much about problem domain “taste” and there is no one correct answer (that is token cost efficient).
Does the smartest agent need the full flavor of the context of the problem to make real progress (i.e what’s been tried, how you’ve influenced the approach) or does the smartest agent really need a crisp problem definition and context room to really work.
make3 24 minutes ago [-]
I suspect that a lot of the fudging around with MCPs and tools can be delegated to smaller agents. Astra doesn't need to look at 300 websearch results directly or to extract text out of raw HTML
koolba 2 hours ago [-]
> My theory is that if you make Qwen understand you, Qwen can elaborate as much as needed to get any of those models get good results, because it's filling in details.
How is this different than just telling the model you’re using to fill in the gaps itself?
If you’re not manually reviewing the intermediate hop, how does that add to the “understanding you”?
xlayn 2 hours ago [-]
there is no difference... each case may be different, I have a local qwen gpu setup... and is fast... but not as fast as claude.
Fable is amazing and if it cost the same as my qwen setup I'll be running fable and never had qwen in first place.
But Fable is crazy expensive, Qwen is not as fast but cheaper... Haiku is fucking dumb IN MY HANDS... BUT my local qwen is quick enough to be briefed, get the idea and delegate.
I work with Fable mostly delegating to opus/sonnet and it's cheaper than just fable, I can't stand working with opus or the other models.
Why Qwen directing the others?
For the things claude refuse, Qwen can make very specific questions and still get an answer without triggering the thing.
make3 26 minutes ago [-]
I think that his idea is that a lot of preparation / setup work is boiler plate / brain dead stuff that can be done by a shallower model.
It's like, find the relevant scientific papers by using Google search and the scientific paper MCPs, summarizing stuff, etc. Then the big model can come in and actually think on the artifacts
ninininino 2 hours ago [-]
Anthropic has a native solution to this which is that you can use `/advisor` to adjust config to enable using Sonnet but set Opus as an advisor or use Haiku and then set Sonnet or Opus as advisor that the weaker model will automatically call when it thinks something warrants it:
It allegedly will automatically judge when to call the stronger model, not sure how much instructions in prompts or AGENTS.md can steer it.
jimbokun 58 minutes ago [-]
Add this to the things I never imagined when thinking about what a world with artificial intelligence might look like.
nico 45 minutes ago [-]
This is very cool. But also, it’s a benchmark/result for a very specific experiment/task: identity resolution
Despite the title, the article is not making claims of, or demonstrating general usage improvements or general coding agent improvements
theanomaly 27 minutes ago [-]
Agreed, though many kinds of tasks are likely to benefit from this pattern, YMMV.
Fordec 2 hours ago [-]
This sort of stuff has me believing that decision model controlled/prioritized context is the future of context management beyond the raw AGENTS.md file.
Why give all the context to an agent, when the context can also be curated, updated and made concise automatically for each new problem.
bigcat12345678 2 hours ago [-]
Yes × 1000...
Jev is the classic type of scientific discovery, any human can see it's effectiveness without actually trying it. We were only deprived the discovery because of the incredible noises created by the AGI/ASI ferver....
gbrindisi 2 hours ago [-]
Excellent read. I use the same hybrid approach for a lot of security engineering work and it’s working well after initial calibrations, for example for our own harness we rebuilt auto mode exactly like this to judge tool calls and intent.
morsch 2 hours ago [-]
I had the same idea, I extended opencode-warden to use a local decision model to validate tool inputs and outputs. It seems to work... ok? d1 is fast enough to run locally on an old 1080 GTX. Not sure if I can recommend relying on it.
boutell 1 hours ago [-]
Ideally, "system one" intelligence is good both at quick reactions and at knowing when to hold off and wait for "system two." Sure enough, that works with Jev.
I don't usually get excited about the high-level, "whoooa man" marketing-speak, but the parallel is striking.
dangoodmanUT 1 hours ago [-]
We've been doing this for a while. Fine tuning the decision model helps a lot too. The problem is you still need a pretty low confidence threshold (e.g 10%), but often times that reduces the work by 70% with 0% quality loss (in our benchmarks)
ltsSmitty 2 hours ago [-]
Liked the read. I wish there were some code examples for how jev/Luna were actually done as I try wrapping my head around their new paradigm.
hadlock 1 hours ago [-]
One of the most immediate and practical uses for AI for most (all?) businesses is CRM enrichment; this is a great article, thanks for sharing.
Jaxkr 2 hours ago [-]
> We tested whether fast, cheap decision models like Jev and GPT-6 Luna
GPT-6 Luna is a small LLM, not a decision model :P
I wonder if something like this is the undisclosed new approach from Anthropic that presumably has massive improvements by combining multiple models. Leveraging decision models as much as possible seems logical for extra speed and token efficiency.
couldaidothis 1 hours ago [-]
I think OpenAI and Anthropic have both been doing similar "routing" approaches for a while. Wasn't OpenAI one of the first to have a small model start responding to you for the first several tokens so that you got an instant response while the big model became available?
Similarly, Grok has been routing to Opus internally for a little while now I think.
2001zhaozhao 2 hours ago [-]
> undisclosed new approach from Anthropic
Any source on that?
pearcoder 2 hours ago [-]
they do have the 'advisor' model setting now which also plays well with this
intervolz 2 hours ago [-]
4x lower is so mathematically awkward.
markcerqueira 1 hours ago [-]
That reads fine to me as a native-ish English speaker. What would read less awkward for you?
mjuarez 1 hours ago [-]
Like a comment on another thread, 0.25x would be the much better way to say this. It's 25% of the original value. It's also consistent: If 3x means 300% the original value, then 0.25x is 25% of the original value.
If you say 4x lower, that changes the semantics for lower than x vs higher than x, which is absurd.
ant6n 2 hours ago [-]
So like, -3x?
timClicks 2 hours ago [-]
More like 0.25x.
xfalcox 1 hours ago [-]
Been doing this on Discourse for Spam and locale detection, mostly doing d3-turbo with Qwen 3.8 Flash Next Help.
phendrenad2 1 hours ago [-]
Jev is interesting because it adds "muscle memory" to LLMs. I think we're approaching some fusion of systems that approaches the human brain here.
giancarlostoro 2 hours ago [-]
What about Sonnet? Haiku? Does it need Opus specifically?
rickypp 2 hours ago [-]
Seems to me that the important factor is that Jev has well-calibrated confidence for this task. Which secondary expensive model it defers to is probably less critical in this technique as long as that model has higher performance than Jev.
couldaidothis 1 hours ago [-]
Yeah exactly - very interesting that Luna is calibrated so poorly despite working better on easier problems.
asar 2 hours ago [-]
i'm so numb to these headlines nowadays... i guess i've been disappointed by how effective these approaches end up being for actual work but i feel like it's easy to produce these headlines with "proof" that doesnt prove anything besides the sandboxed scenario it was run in
Update: let me elaborate more, I read the HN entry about the team of 5 Qwens being close to fable quality... so I embarked to add to my harness (https://github.com/alainnothere/privibe because we all have one right?) and once I had added the orchestrate tool I had to now measure somehow it it's able to improve quality, so I claude and then Qwen to run a experiment to call my harness passing the model to use and ask them to do something, 3 rounds I think and see what quality came out of it... and as per the grading... qwen and claude agreed that the quality of the response comming out the harness using the orchestrate was VERY CLOSE, with opus and fable providing more "in my opinion you should" which matches my experience on why they are better.
One more update: I can't get copilot models to work decently, it may be my way of working... but I can get Qwen to follow what I try to do... if Qwen get's me he can then fire copilot cli and get amazing results by using copilot... bonus points seems like copilot cli is cheaper with the AIC (cost of the requests) so you can make the copilot subscription longer
https://opencode.ai/docs/agents/
There are a lot of ways to do this if you stop using Big Ai harnesses that make it difficult to use models besides theirs.
My sense so far is it’s very much about problem domain “taste” and there is no one correct answer (that is token cost efficient).
Does the smartest agent need the full flavor of the context of the problem to make real progress (i.e what’s been tried, how you’ve influenced the approach) or does the smartest agent really need a crisp problem definition and context room to really work.
How is this different than just telling the model you’re using to fill in the gaps itself?
If you’re not manually reviewing the intermediate hop, how does that add to the “understanding you”?
Why Qwen directing the others? For the things claude refuse, Qwen can make very specific questions and still get an answer without triggering the thing.
It's like, find the relevant scientific papers by using Google search and the scientific paper MCPs, summarizing stuff, etc. Then the big model can come in and actually think on the artifacts
https://code.claude.com/docs/en/advisor
It allegedly will automatically judge when to call the stronger model, not sure how much instructions in prompts or AGENTS.md can steer it.
Despite the title, the article is not making claims of, or demonstrating general usage improvements or general coding agent improvements
Why give all the context to an agent, when the context can also be curated, updated and made concise automatically for each new problem.
I don't usually get excited about the high-level, "whoooa man" marketing-speak, but the parallel is striking.
GPT-6 Luna is a small LLM, not a decision model :P
Similarly, Grok has been routing to Opus internally for a little while now I think.
Any source on that?
If you say 4x lower, that changes the semantics for lower than x vs higher than x, which is absurd.