Kimi-K3 vs Cosmos-Reason2-2B
Specs, VRAM requirements and download trends — updated 21 August 2026.
Specification comparison
Differences are highlighted; identical values are muted.
| Specification | Kimi-K3 | Cosmos-Reason2-2B |
|---|---|---|
| Deployment | Open weights · also via API | Open weights |
| Parameters | 2,779.9 B | 2.4 B |
| Architecture | KimiK3ForConditionalGeneration | Qwen3VLForConditionalGeneration |
| Context length | 1,048,576 | — |
| Licence | other | other |
| Commercial use | Unknown | Unknown |
| Languages | — | — |
| Downloads 30d | 922,082 | 840,398 |
| Downloads all time | 5.7 M | 3.3 M |
| Quantizations on the Hub | SAFETENSORS | SAFETENSORS |
| Gated | No | Yes — accept licence |
| First seen on the Hub | 2026-06-13 | 2025-12-12 |
| API input price / 1M tokens | $0.24 | not on OpenRouter |
| API output price / 1M tokens | $10.80 | not on OpenRouter |
| Est. cost per 1M input + 1M output tokens | $11.04 | not on OpenRouter |
| Cached input / 1M tokens | $0.21 | not on OpenRouter |
| AA Intelligence Index | 43.6 | not on OpenRouter |
| AA Coding Index | 76.2 | not on OpenRouter |
| AA Agentic Index | 50 | not on OpenRouter |
| Max output tokens | 943,718 | not on OpenRouter |
| API providers | 21 | not on OpenRouter |
| Tool calling | Yes | not on OpenRouter |
| Reasoning mode | Yes | not on OpenRouter |
API cost
Estimated cost of 1M input + 1M output tokens via OpenRouter
Kimi-K3
$11.04
Download trend
Daily snapshots, last 82 days (4 Aug – 24 Oct)
2.9 M831 K
Kimi-K3
Cosmos-Reason2-2B
VRAM side by side
On RTX 4090 · 24 GB · 8K context unless noted
Cosmos-Reason2-2B · fp16
6.2 GB / 24 GB
Kimi-K3 · fp16
2,134.5 GB / 24 GB