$0.39 per 1M input tokens and $1.63 per 1M output tokens (cheapest host: Featherless).
There is no free API; the cheapest paid price is $0.39 per 1M input tokens. The weights are open, so you can run it yourself.
#141 of 334 in Chat & Knowledge, #105 of 239 in Coding, #132 of 242 in Agents & Tool Use, #79 of 195 in Writing & Language.
On Aider Polyglot · two attempts, R1 0528 scores 71.4 vs 81.3 — lower.
Through the API of DeepInfra, Featherless, Novita, SiliconFlow, StreamLake.
128K tokens, with up to 33K output tokens.
Released 2025-05-28. It is the current version in its line.
Yes, the weights are open (MIT licence).
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
No, it accepts text only.
R1 0528 is an open-weights language model from DeepSeek, released May 2025.
GPT-5 (better), Kimi K2 0711 (open weights).
BetterGPT-5high+23%+395%$1.25 / $10
OpenKimi K2 0711unstated-17%+26%$0.50 / $2Memory (4-bit): 411 GB → 616 GB+1
Qwen3 32BunstatedOpen-44%+76%$0.70 / $2.80Context: 128K → 33K