AI & Tech Developments - Apr 04

New benchmarks show Gemma 4 outperforms Qwen 3.5 in coding tasks.

— @leftcurvedev_

New method for comparing AI models introduced by Anthropic.

— @AnthropicAI

Claude subscriptions no longer cover usage on third-party tools.

— @bcherny

Gemma 4-31B scores better than several models in benchmarks.

— @TeksEdge

China rejects Israeli request to criminalize anti-Israel sentiment.

— @MENA_Puls

Qwen 3.6 Plus breaks 1 trillion tokens processed in a single day.

— @OpenRouter

GPT-Image-2 testing on Arena shows mind-boggling results.

— @blakeir

Gemma 4 31B is the best open dense model on the market.

— @teortaxesTex

Claude's token usage significantly reduced by caveman mode.

— @om_patel5

New AI agents can be built without technical skills.

— @AWSstartups

New benchmarks show Gemma 4 outperforms Qwen 3.5 in coding tasks.

— @leftcurvedev_

New method for comparing AI models introduced by Anthropic.

— @AnthropicAI

Claude subscriptions no longer cover usage on third-party tools.

— @bcherny

Gemma 4-31B scores better than several models in benchmarks.

— @TeksEdge

Qwen 3.6 Plus breaks 1 trillion tokens processed in a single day.

— @OpenRouter

GPT-Image-2 testing on Arena shows mind-boggling results.

— @blakeir

Gemma 4 31B is the best open dense model on the market.

— @teortaxesTex

Claude's token usage significantly reduced by caveman mode.

— @om_patel5

New AI agents can be built without technical skills.

— @AWSstartups

Commentary on the efficiency of energy production from different sources.

— @vbenard

Testing bartowski/google_gemma-4-26B-A4B-it-GGUF IQ4_NL with Hermes Agent shows speed and accuracy.

— @JoelDeTeves

MIT formally proved that ChatGPT is structurally trained to make you delusional, and even a perfectly rational user can spiral into delusion in the long term. Knowing about it doesn't save you either.

— @Hesamation

Qwen3.5 9B rodando em 6.3GB de VRAM. 42.8 tok/s. Cabe em QUALQUER GPU de 8GB pra cima. RTX 4060, RTX 3060, T4. Tudo roda. KV cache com PolarQuant Q3: 5.3x de compressao sem perder velocidade (43.4 tok/s comprimido vs 42.8 sem).

— @0xCVYH

The ultimate Gemma-4-31b-UNCENSORED

— @dealignai