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.
Moonshot's[3] tech blog says "We are open sourcing our latest model, Kimi K2.6" and "is now available"; the Hugging Face[2] README went public on 2026-04-20, OpenRouter[4] listed kimi-k2.6 the same day and NYU's[6] write-up states "released Kimi K2.6 on April 20, 2026", while DataNorth[5] gives April 21.
Moonshot's[3] model card: 1T total and 32B activated parameters, 384 experts with 8 selected plus 1 shared, 256K context, 400M-parameter MoonViT, Modified MIT license and the benchmark table against GPT-5.4, Claude Opus 4.6, Gemini 3.1 Pro and K2.5.
Moonshot's current API price table lists kimi[1]-k2.6 at $0.16 per million cached input tokens, $0.95 input and $4.00 output with a 262,144-token context, and shows K2.5 discontinued on 31 August 2026.
OpenRouter's record for kimi[1]-k2.6: created 2026-04-20, 262,144-token context, text and image input, and third-party prices of about $0.43 to $1.09 per million input tokens and $1.82 to $4.60 output across 17 providers.
22 April write-up: benchmark highlights, availability through the Moonshot[3] API, Ollama, Cloudflare Workers AI and OpenRouter[4], and launch pricing of $0.60 input, $4.00 output and $0.16 cached per million tokens.
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.