qwenBerbayar
Qwen: Qwen3 235B A22B Thinking 2507
qwen/qwen3-235b-a22b-thinking-2507
Qwen3-235B-A22B-Thinking-2507 is a high-performance, open-weight Mixture-of-Experts (MoE) language model optimized for complex reasoning tasks. It activates 22B of its 235B parameters per forward pass and natively supports up to 262,144...
Harga input / 1M token
$0.2300
Harga output / 1M token
$2.30
Context length
131.072
Modalitas
- Input
- text
- Output
- text
Parameter yang didukung
frequency_penaltyinclude_reasoninglogprobsmax_tokenspresence_penaltyreasoningrepetition_penaltyresponse_formatseedstoptemperaturetool_choicetoolstop_ktop_logprobstop_p
Rilis
25 Juli 2025