Haven't added a demo to the repo yet (will do soon) but building this TUI [1] in Go to manage my agent skills.
I don't like putting 20-30 agent skills in the .claude/skills/ dir and letting the agent figure it out. I keep all the skills I've created and adapted over time in a separate git repo and then from there I copy them to the project skill dir when i need them for a session and then remove them when I'm done.
This TUI just replaces all the manual work with ls, cp -r, and rm -rf commands with a few keystrokes.
I’d just like to add, and this may interest you both, that I’m not disagreeing with either of you. When 3.5 Flash first showed up in AI Studio, its model card said it had a March 2026 knowledge cutoff. After the backlash over it apparently knowing nothing past December 2024, the card was changed to read: “Knowledge cutoff: Unknown.” Maybe the timing was coincidental, but I doubt it.
I guess they meant to release Gemini 3.5 Pro shortly after 3.5 Flash, but then Mythos/Fable and later GPT-5.6 came out with higher performance than 3.5 Pro, so the managers decided not to release it.
3.6 Flash would be a great model at 3.0 flash pricing. At this pricing, its thoroughly trounced by about 10 models on cost/performance including Grok 4.5.
3.5 Flash-ite would be a great model at 2.5 flash-lite pricing, as is, its trounced by many models including Deepseek v4 Flash.
As is, they are thoroughly outclassed for most usecases. I will say the one area where i do see Gemini punching above its weight class is in tasks that are effectively "Google this for me" / knowledge stuff. So it does have a role, and I do use it. So while I think Google is still in a strong position overall, they are really stuck as a tier 2 AI player right now with text models. They are tier 1 in bio, images, and video.
Could also be that they are pricing it at levels where they actually make money. Without seeing the behind the scenes compute cost on all of these its hard to really judge.
That being said with any open model we of course do know the total cost (or estimate)
3.5 Flash was always too expensive for a "flash" model. They marketed it as "near frontier" level, but there are several order-of-magnitude cheaper open models that compete with it.
In my tests, 3.6 Flash is NOT more token efficient, so it actually ends up costing more than 3.5 Flash, even with the output price reduction.
EDIT:
It less less verbose in final output though, but it reasons more.
I assume the optimization comes when you have long-running tasks with many tool calls, and by reasoning more, it reduces the number of tool calls needed.
It seems like they're sticking to a static pricing plan that was made when the only relevant competition were the US labs (im not counting deepseek 2025 as serious competition -> glm and then kimi on the other hand, now that's a different story).
Pricing often reflects what the vendors (expects) the customer is willing to pay. It seems that Google is still trying to find their niche in the market.
I wonder if this is a plateau towards the real pricing of AI, if you layer in gemini-2.0-flash at $0.10 / $0.70 then its a 15x price increase to 3.5/3.6 flash. But it hasnt gone up again which is interesting.
Handy is an amazing cross-platform app for dictation from the author. There are other awesome open-source dictation tools as well like native macOS ones. You do not need SaaS subscription in this day and age for transcription.
I maintain this list of all the best open-source ones in this awesome-style GitHub repo. People looking for open-source dictation tools, hope you find something that works for you here:
If you're talking about translated text, then that should be super easy. Most of these dictation tool support post-processing with LLM to remove filler words, fix punctuation, etc. I'd imagine you can change the system prompt for the post-processing step to do the translation instead, and you'd get translated text.
Yea I’m looking local hosted transcription and translation with diarization of 2 (or more ) speakers. This is to speed up collaborative technical work between two teams who speak different languages where want all local processing (assume no cloud access).
Totally understandable, but I’ve found that software that transcribes everything after I finish recording actually works better for me. I’ve tried both kinds, and systems that continuously type what I’m saying distract me from completing my thought. I end up reading what’s being typed and noticing transcription mistakes instead of focusing on what I’m trying to say.
I often prefer to dictate everything in my head about a particular thing for 5–10 minutes and then go through it afterward. I find that much more useful because it doesn’t break my thought process the way continuous transcription does.
I can understand both modes. I mostly use transcription as input for my AI assistant and there I find it very useful to be able to check my input and just repeat myself in case something wasn’t fully captured. When using Apple’s transcription feature built into iOS and macOS, I also really like being able to edit everything right while the dictation is still active.
Dream would be a combination of AI models that are smart like a human transcriber. One can just tell the model what mistakes it has made (e.g. "No, you wrote XYZ but I meant WXY"), and it is intelligent enough to realize when it is being instructed and when it needs to transcribe exactly what I say.
basic etiquette and presentation (the linked article) is mind blowing enough to reach the top forum ranking, precisely because of behaviour like that which you describe. as the dog whisperer taught me, dominance and fear are seperate concepts - it is possible to set boundaries without being a dick. personally I love being a dick but moderation is key. snarky behaviour shifts responsibility to avoid directly addressing the issue oneself. thats not how an adult behaves in my book.
I don't like putting 20-30 agent skills in the .claude/skills/ dir and letting the agent figure it out. I keep all the skills I've created and adapted over time in a separate git repo and then from there I copy them to the project skill dir when i need them for a session and then remove them when I'm done.
This TUI just replaces all the manual work with ls, cp -r, and rm -rf commands with a few keystrokes.
[1] https://github.com/primaprashant/sei
reply