Quantization's Hidden Tax, Cliff Tokens, Smarter Memory
Three new arXiv papers reveal hidden costs in quantized reasoning models, single-token failure triggers, and a new framework that cuts agent memory errors by up to 79%.
Articles Tagged "Reasoning"
AI Research: Orchestration Beats Scale, Small Models Win
Sakana Fugu tops SWE-Bench Pro by routing tasks across rival LLMs, Microsoft's 9B browser agent beats OpenAI Operator, and a 3B model from Weibo matches DeepSeek V3.2 on math.
VibeThinker-3B
WeiboAI's 3B dense reasoning model fine-tuned from Qwen2.5-Coder-3B, posting AIME 2026 scores that match DeepSeek V3.2 (671B) using the Spectrum-to-Signal training pipeline.
ERNIE 5.1
Baidu's ERNIE 5.1 is a text-focused MoE model that claims the top Chinese model slot on LMArena with 800B parameters built at 6% of comparable training costs.
Emergent Alignment, Agent Memory, and Smarter Reasoning
Three arXiv papers: a conscience mechanism for ethical training, shared memory for agent populations, and selective verification that cuts test-time compute waste.
Pramaana Labs Raises $27M for Provably Correct AI
Pramaana Labs uses the LEAN proof language to attach a mathematical certificate to every AI answer in high-stakes domains like tax, law, and drug discovery.
GPT-5.1
GPT-5.1 is OpenAI's November 2025 coding and agentic flagship with 400K context, configurable reasoning effort, and 76.3% on SWE-bench Verified.
Kimi K2.7-Code
Moonshot AI's Kimi K2.7-Code is a 1T-parameter open-weight MoE coding model with mandatory thinking mode, 256K context, and 30% fewer reasoning tokens than K2.6.
Microsoft MAI Models: Seven-Model Suite Reviewed
A hands-on review of all seven MAI models - from the April transcription and image launch to Build 2026's MAI-Thinking-1, MAI-Code-1-Flash, and the multimodal upgrades.
MAI-Thinking-1
Microsoft's first in-house reasoning model, a 35B-active sparse MoE with 256K context, 97% on AIME 2025, and no distillation from third-party labs.
Context Overload, Memory Leaks, and Agent Safety
Three new arXiv papers expose how context bloat tanks agent performance, agent memory bleeds private data, and misaligned behavior spreads through multi-agent systems.
Claude Opus 4.8 Leads SWE-Bench Pro, Adds Parallel Agents
Anthropic's Claude Opus 4.8 scores 69.2% on SWE-bench Pro and ships hundreds of parallel subagents in Claude Code, with pricing unchanged at $5 per million input tokens.
