Insight tag
8 related stories.
Mistral announced Magistral on June 10, 2025 as its reasoning family. A visible rationale still needs source checks, output controls, and accountable decisions.
Read analysisDeepSeek released R1-0528 on May 28, 2025 with stated benchmark, function-calling, and hallucination changes. A reasoning gain can also change tokens, latency, and tool behavior.
Read analysisARC Prize launched ARC-AGI-2 on March 24, 2025, with grid-based reasoning tasks, 120-task evaluation sets, pass@2 scoring, and cost reporting.
Read analysisNVIDIA announced Llama Nemotron reasoning models on March 18, 2025, including Super. A family benchmark needs the exact size, quantization, tools, prompts, and agent harness.
Read analysisQwen released QwQ-32B on March 6, 2025 and described a 32B reasoning model with agent-facing capability claims. Its comparisons need the stated tasks, budgets, and scaffolding.
Read analysisIBM released Granite 3.2 on February 26, 2025 with controllable experimental reasoning and a model card for the 8B instruction artifact, making route policy observable.
Read analysisAnthropic announced Claude 3.7 Sonnet and Claude Code on February 24, 2025. Its hybrid reasoning design made compute limits an operational control.
Read analysisDeepSeek submitted its DeepSeek-R1 paper to arXiv on January 22, 2025. Its account of reinforcement learning and distillation called for claim-by-claim model evaluation.
Read analysis
