On ARC AGI 2, Opus is just a smarter model and I think we see such big climb not because it is so good at thinking, but rather without thinking it is nearly impossible for it to do these puzzles well (same for me). E.g. GPT-5-Pro is below Opus 4.5 thinking 16k, but there's no way
LLMS
-
Voyager SDK 1.5 Launches Ubuntu 24.04 Python 3.12 Support
By
–
So, what’s got you excited? Ubuntu support? Multi-chip power? Or those shiny new CV models? Tell us what you’re building, and get the 1.5 update https://
community.axelera.ai/voyager-sdk-2/
voyager-sdk-v1-5-is-here-with-native-ubuntu-24-04-and-python-3-12-support-1146
… #EdgeAI #VoyagerSDK #Ubuntu2404 #Python312 #LLM #ComputerVision -
LLM Upgrades: Windows Support and New Model Format
By
–
LLM upgrades
• Language models now supported on Windows
• New model format supports future AIPU versions [ Heads-up: Models from older SDK versions need recompiling or redownloading for future-proofing.] -
Metis 4-Chip PCIe Cards Enable Edge AI and LLM Deployments
By
–
New hardware support
Say hello to the new 4-chip Metis PCIe cards (16GB and 64GB).
Ideal for edge deployments needing serious compute: multi-stream apps, larger models, LLMs. -

Voyager SDK v1.5 Released with Ubuntu 24.04 and Python 3.12
By
–
Voyager SDK v1.5 is out now, and it’s a big one.
Native Ubuntu 24.04, Python 3.12, new Metis PCIe support, fresh CV models, and upgraded LLM support. Let’s dig in… -
Claude.ai fixes context limit frustration with automatic compaction
By
–
In case you missed it, earlier this week we fixed one of the most common frustrations on http://
Claude.ai: hitting context limits mid-conversation. Claude now intelligently compacts earlier context automatically when you're nearing the limit so the chat can keep going. -
99% Performance Achievable with Smaller Language Models
By
–
i bet you could get 99% of the performance with less powerful LLMs
-
LLMs Delay Progress: We Need Novel Ideas, Not Incremental Improvements
By
–
they (or more likely more efficient replacements) will play a role. but spending so much money, time, and energy on them has quite likely delayed us. we need new ideas, not slightly better LLMs.
-

Multi-Agent Communication Costs: Beyond Token Overhead
By
–
Multi-agent systems are powerful but expensive. However, the cost isn't in the reasoning itself. It's in the communication. Agents exchange full text messages, consuming tokens for every coordination step. When agents need to collaborate on complex problems, this overhead adds
-
LLM Over-Focus Causing AI Field Stagnation
By
–
you could argue the analogy but the general point is correct; an over focus on LLMs has allowed the field to stagnate.