The first entry is always the pin's source. Overall confidence is a weighted average of how firmly each reference supports the start and end times used above; a reference counts half as much for every 180 days older than the newest.
Cohere's[2][5] blog post "Introducing Command A: Max performance, minimal compute" is dated Mar 13, 2025 and says "Command A is available today on the Cohere platform"; the Cohere release note and OpenRouter's listing carry the same day.
Cohere's[1][5] current model page: $2.50 per 1 million input tokens and $10 per 1 million output tokens, 256,000-token context, 8,000 max output tokens, 23 trained languages, and the Command A Reasoning, Vision, Translate and A+ siblings in the nav.
Model card: 111B parameters, CC-BY-NC 4.0 with Cohere[1][2][5] Labs' acceptable use policy, 256K context configured at 128K in Hugging Face, three sliding-window layers (4096) per global layer, GPQA Diamond 50.51, gated access.
Cohere's[1][2][5] technical report, submitted 1 April 2025: hybrid architecture, decentralised training with self-refinement and model merging, 23 languages, weights for Command A and Command R7B released for research.
Cohere's[1][2] release note dated March 13, 2025: 111B parameters, 256K context, 150% higher throughput than Command R+ 08-2024, two A100 or H100 GPUs, available today on the Cohere Platform and Hugging Face[3] as command-a-03-2025.
Something missing or wrong? Say it in your own words: a link that backs this pin up, a different start or end date and why, or a fact it lacks or gets wrong. The AI checks it against this pin's sources, searches for better ones, and adds any page that backs you up. The pin's own sources still count most. A picture that shows something else, or shows it badly, is looked at too, and moved down or replaced.