Yes, V4.1 Flash 1017 is my daily driver cloud model. Copes fine with what I'd consider difficult brownfield development, like fixing bugs in vLLM prompt cache and coming up with small/sane overlay patches for them. Maybe it's disappointing for people who want to one-shot a game or website from a single prompt; I haven't tried that.
It's nice that their peak-rate hours end quite early in the day UK time.
You can do real world development with weaker models than Deepseek Flash.
I still use SOTA for the main model though, but also I'm perfectly aware it's mostly out of FOMO.
Deepseek is more than enough for default subagents though. Any well defined and verifiable task and there's absolutely zero difference between SOTA and good-enough models. Used to be just speed, but Deepseek is fast.
I have same the same FOMO to use SOTA model for all planning and brainstorming session. However, I’m unsure how do you bring deepseek for subagent task. Could you describe your workflow?
For myself, using Codex, I always have to explicitly tell codex to use Luna with high reasoning for smaller, bit size tasks.
I'm using OMP (oh-my-pi) and the minimum you need to do is just set the @task role to Deepseek:max.
Subagents can spawn with other profiles e.g. scout (uses @smol) and reviewer (@slow). You can set SOTA with higher thinking for @slow and Deepseek with lower thinking for @smol or something even faster/cheaper.
There's also weirdly named /vibe mode that gives explicit instructions to the main model to act as an orchestrator but I find this unnecessary if your main is SOTA, they are already super eager to spawn subagents by default.
Yes, V4.1 Flash 1017 is my daily driver cloud model. Copes fine with what I'd consider difficult brownfield development, like fixing bugs in vLLM prompt cache and coming up with small/sane overlay patches for them. Maybe it's disappointing for people who want to one-shot a game or website from a single prompt; I haven't tried that.
It's nice that their peak-rate hours end quite early in the day UK time.
I use it at work. It's dirt cheap, blazing fast, and is more than enough for most coding work.
If you live in a country with a weak currency, DeepSeek models are definitely a godsend.
You can do real world development with weaker models than Deepseek Flash.
I still use SOTA for the main model though, but also I'm perfectly aware it's mostly out of FOMO.
Deepseek is more than enough for default subagents though. Any well defined and verifiable task and there's absolutely zero difference between SOTA and good-enough models. Used to be just speed, but Deepseek is fast.
I have same the same FOMO to use SOTA model for all planning and brainstorming session. However, I’m unsure how do you bring deepseek for subagent task. Could you describe your workflow?
For myself, using Codex, I always have to explicitly tell codex to use Luna with high reasoning for smaller, bit size tasks.
I'm using OMP (oh-my-pi) and the minimum you need to do is just set the @task role to Deepseek:max.
Subagents can spawn with other profiles e.g. scout (uses @smol) and reviewer (@slow). You can set SOTA with higher thinking for @slow and Deepseek with lower thinking for @smol or something even faster/cheaper.
There's also weirdly named /vibe mode that gives explicit instructions to the main model to act as an orchestrator but I find this unnecessary if your main is SOTA, they are already super eager to spawn subagents by default.