no. gpt is just the better software engineer.
LLMS
-
Claude vs Codex: Different Strengths for Different Tasks
By
–
I don't see a conflict there. Claude is better for the lolz
codex is the better software engineer For an agent there's many that prefer lolz -
Benchmarking AI Agents with 26B 8-bit Model on Jetson Orin
By
–
Justement là je passe toute la soirée à le tester sur de l’agentique avec openclaw + benchmark sur la jetson nano orin. Moi le chiffre qui m’intéresse le plus c’est le 166 sur 26B 8-bits c’est juste hallucinant je trouve d’en être déjà arrivé là.
-

Benchmarking Gemma 4 and 26B AI Models on NVIDIA RTX 5090
By
–
Pour ceux que ça intéresse, j’ai lancé un benchmark sur ma NVIDIA RTX 5090 avec une utilisation "normale" de l'ordinateur en parallele sur Gemma 4. Je suis à 285.29 tokens par seconde sur Gemma 4 E2B en 4 bits. Quelle folie quand même. Et le 26B en 8 bits tourne à 166.75
-

Unsloth AI launches MLX Dynamic Quants for Gemma-4
By
–
Wait, what? @UnslothAI is starting to upload MLX Dynamic Quants! I have to test them ASAP! Thanks you really rock! 🚀 unsloth.ai/docs/models/gemma…
→ View original post on X — @huggingface, 2026-04-04 19:48 UTC
-

Claude Opus 4.6 Independently Discovers Benchmark Answers
By
–
Anthropic revealed that its Claude Opus 4.6 model, when subjected to an evaluation test, spontaneously identified that it was taking an exam, located the benchmark's GitHub repository, and decrypted the expected answers. Eighteen times. No one had asked it to. My article with @dr_l_alexandre [Translated from EN to English]
→ View original post on X — @alex_tsico, 2026-04-04 19:27 UTC
-
Chain of Thought as Context Compaction in AI Models
By
–
Interesting.. Chain of thought is a reduce (in addition to attention ofc), so I guess this can be seen as a bit more of a directed context compaction mechanism, inheriting structure from the preexisting idea of a wiki.
-
Switching from Third-Party to First-Party Claude Model Harness
By
–
With this you are not using a 3rd-party harness anymore, but the 1st-party claude one. I recommend switching to a more friendly model provider tho. There's quite a lot that offer good subs.
-
Claude -p harness call allowed by Anthropic
By
–
this calls claude -p, which is their 1P harness and allowed.
-

OpenAI Preparing Imminent Release of Image Generation Model v2
By
–
OpenAI’s image gen v2 release immanent