and Long live Wikipedia!

Çeviri için teşekkürler, orijinal post: https://lemmy.zip/post/3862

repost in case if imgur is down

yeah I do I even use diff tools to see if they hallucinate

Still same, I writed a post that explains why they suck https://lemmy.zip/post/58970686

uBlock Origin already fixed that https://lemmy.zip/post/57524806/24557587

The real reason is LLMs are still using the same architecture and there is no breakthrough at the end of the day their intelligence will become so close to each other, when this happens they will have to decrease the prices to compete with open-weight models and even with these prices they don’t generate revenue so instead of just scaling they will have to focus on optimization and innovation

Subscriptions is stupid. I pay for tokens and I’m not lock-in one provider

right now you can use a Qwen-3-4B fine tuned model (Jan-v1-4B) with search tool and get even better results than Perplexity Pro, and this was 6 moths ago

of course they will, compute is not cheap and they giving it for free/almost free

This has nothing to do with Markdown. It’s disinformation from Microslop.

You can make the link C:\windows\system32\cmd.exe hn

This is so stupid. Why did they add something like this? In Markdown, there is no execution. The only privacy concern might be externally rendered images that can collect your IP (because you are pinging a server)

Free Software is essential

Filterlist mainteiners are GOAT

Besides, clothing without logos is more expensive than others.

GLM-5 and Kimi-K2.5 is really good.

ArtificialAnalysis Intellegence vs Cost

yes GPS accuracy is better but it slower

OpenRouter is basically the place for third-party AI stuff: tools, research, benchmarks…

Sure, it won’t tell you what regular users are doing, but it shows where professionals actually spend money. And since it’s pricey (seriously, once you’re used to free ChatGPT or Gemini, paying hurts :D), it reveals which models are actually worth it.

microG allows choosing any location server /v1/geolocate , you can even import OpenCell data and use fast location services completely offline

General ranking (weekly) (higlighted models are open-weight)

General ranking (weekly)

1. Kimi K2.5 - 1.45T tokens
2. Gemini 3 Flash Preview - 737B tokens
3. DeepSeek V3.2 - 711B tokens
4. Claude Sonnet 4.5 - 678B tokens
5. MiniMax M2.1 - 454B tokens
6. Gemini 2.5 Flash - 449B tokens
7. Grok 4.1 Fast - 421B tokens
8. Trinity Large Preview - 388B tokens
9. Gemini 2.5 Flash Lite - 358B tokens
10. Claude Opus 4.5 - 345B tokens
11. Grok Code Fast 1 - 314B tokens
12. Claude Opus 4.6 - 275B tokens
13. gpt-oss-120b - 266B tokens
14. GPT-5 Nano - 265B tokens
15. Gemini 2.0 Flash - 175B tokens
16. GLM 4.7 - 171B tokens
17. Gemini 3 Pro Preview - 169B tokens
18. Pony Alpha (GLM-5) - 147B tokens
19. GPT-5.2 - 145B tokens
20. Claude Haiku 4.5 - 132B tokens

Wow 46% of tokens are now going through open-weight models thats amazing.

yes and GPS is very slower than WiFi and Cell Towers