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.
I use it through my opencode go subscription and it's exactly how you described. Very pragmatic and not too ambitious. It's similar to Kimi 2.5/6 in that regard.
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)
No two uses are alike, I suppose. For me, whatever difference is a wash. However, I probably tend to shy away from throwing high-complexity/long-horizon tasks at the model.
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 have Opus make a fairly detailed plan, then Deepseek implements, and GPT reviews. With that setup, I have zero issues, probably because what you mention is handled (the plan keeps it on track and the reviewer catches any issues).
Now that you mention it, though, I have seen it do a few things that weren't in the plan. The reviewer caught them, though, so they didn't cause a problem, and it's so cheap that overall it's a massive improvement.
It's the only model where an explicit instruction at the end of my message is sometimes ignored. This doesn't happen with any of the gpts, kimis, glms, qwen, etc. Just a deepseek problem.
I have also noticed this with Sonnet, funnily enough - it's not as strong, but it's still there. But yeah, I haven't seen this with any other model so far (although I mostly use the stronger ones - maybe it's a function of intelligence?).
Cursor with Composer 2.5 seems to be competitive with frontier models (Opus and GPT-5.5) for a significant price discount. Benchmarks are gamed, as always, but $0.55/task vs $11.02 a task definitely indicates that there's some cost advantage.
If you want to understand why Alberta is holding a referendum on whether they should hold another separate legally-binding referendum in the future, you have to look at the recent court case where a judge in Alberta ruled that one of the two main petitions wasn’t allowed to proceed (The one that specifically called for a legally-binding referendum). The judges stated reason is that First Nations were not adequately consulted (interesting how this never came up in the Quebec referendums). As a result, the premier of Alberta suggested that until they appeal that court case that they cannot have a legally binding referendum. As such, for now, all they cannot do is a non-legally binding referendum on whether they should hold a legally binding referendum once they court case becomes resolved.
"Consistent with this long tradition of respect for minorities, which is at least as old as Canada itself, the framers of the Constitution Act, 1982 included in s. 35 explicit protection for existing aboriginal and treaty rights, and in s. 25, a non-derogation clause in favour of the rights of aboriginal peoples. The "promise" of s. 35, as it was termed in R. v. Sparrow, [1990] 1 S.C.R. 1075, at p. 1083, recognized not only the ancient occupation of land by aboriginal peoples, but their contribution to the building of Canada, and the special commitments made to them by successive governments. The protection of these rights, so recently and arduously achieved, whether looked at in their own right or as part of the larger concern with minorities, reflects an important underlying constitutional value."
95% of Alberta is unceded First Nations land. It is not a valid country without it - without the consent of the relevant First Nations, a separated Alberta would be a few municipalities enveloped by... Canada.
This is not a concern in Quebec, because the overwhelming majority of it is ceded land.
If ducks had two wheels, they'd be bicycles, and if there was anything in common between the two provinces, you might have a point.
As far as I know, First Nations lands in Alberta are indeed ''ceded'' under Treaties 6, 7, and 8 with the Crown. British Columbia is a province with a huge proportion of unceded land, but not Alberta.
The treaties with the crown require the crown consult with them before adjusting them. This means that Albertan secession can't happen without their consent, as it would by definition, completely and unilaterally adjust the terms of those treaties.
The treaties were made in perpetuity, and if you are going to not hold up the crown's end of the promises, the FN's side - giving the crown and Alberta governance over the land - needs to be reverted as well.
So because Quebec ancestors killed all the people who opposed the conquest of that land, it's okay for Quebec to secede?
But because another set of Canadians didn't kill off all the natives that still claimed Alberta's land, they can't secede legally?
Is that the logic?
Wow, I'm not sure where you're going with that. Read up on the ''Brandy Parlit'' debates and you'll see that genocide of indigenous people was never at play in early Quebec. The relationship between European colonists in New France/Lower Canada/Quebec and First Nations has always been frought, but not genocidal.
you might want to do some amount of research before accusing the french ...
who by far out of the colonial powers had the best relationship with indigenous people in north america, and whos relationships created a new culture blending french and indigenous cultures together into the Metis,
I wouldn't say that. There's plenty of behaviors where plants and fungi act in the interest of the community as a whole. Mature forests are made of late-successional species. These species could only possibly grow after earlier succession species had "prepped the soil" (nurtured the soil ecology, transitioned the nitrogen cycle from ammonia to nitrates, etc) and provided shade to allow them to grow. Since these trees are dependent on the well being of so many other species there is plenty of incentive for them to act in ways that we might call "selfless"
Which is a far cry away from the “free market” we see in the west.
Wild how I got downvoted for stating a pretty obvious fact. Long term planning and self sacrifice in the interest of long term community benefit which transfers to one’s future self / progeny is actually very similar to communism.
reply