qwen/qwen3-235b-a22b-instruct-2507

Updated Qwen3 model for instruction following

Capabilities

Max TokensTop-P

Community model (estimated from hardware time)

Name	Type	Description	Default	Constraints
`frequency_penalty`	number	Frequency penalty	`0`	min: -2, max: 2
`max_tokens`	integer	The maximum number of tokens the model should generate as output.	`1024`	min: 1, max: 16384
`presence_penalty`	number	Presence penalty	`0`	min: -2, max: 2
`prompt`	string	Prompt	`""`	—
`temperature`	number	The value used to modulate the next token probabilities.	`0.1`	min: 0, max: 2
`top_p`	number	Top-p (nucleus) sampling	`1`	min: 0, max: 1

frequency_penaltynumber

Frequency penalty

Default: 0min: -2, max: 2

max_tokensinteger

The maximum number of tokens the model should generate as output.

Default: 1024min: 1, max: 16384

presence_penaltynumber

Presence penalty

Default: 0min: -2, max: 2

promptstring

Prompt

Default: ""

temperaturenumber

The value used to modulate the next token probabilities.

Default: 0.1min: 0, max: 2

top_pnumber

Top-p (nucleus) sampling

Default: 1min: 0, max: 1

Version: c619bc062df6Updated: 7/25/20261.1M runs