Qwen: Qwen3 235B A22B Instruct 2507

Text input Text output
Author's Description

Qwen3-235B-A22B-Instruct-2507 is a multilingual, instruction-tuned mixture-of-experts language model based on the Qwen3-235B architecture, with 22B active parameters per forward pass. It is optimized for general-purpose text generation, including instruction following, logical reasoning, math, code, and tool usage. The model supports a native 262K context length and does not implement "thinking mode" (<think> blocks). Compared to its base variant, this version delivers significant gains in knowledge coverage, long-context reasoning, coding benchmarks, and alignment with open-ended tasks. It is particularly strong on multilingual understanding, math reasoning (e.g., AIME, HMMT), and alignment evaluations like Arena-Hard and WritingBench.

Key Specifications
Cost
$$$
Context
262K
Parameters
235B
Released
Jul 21, 2025
Speed
Ability
Reliability
Supported Parameters

This model supports the following parameters:

Presence Penalty Stop Top P Tool Choice Temperature Min P Seed Tools Frequency Penalty Max Tokens
Features

This model supports the following features:

Tools
Performance Summary

Qwen3-235B-A22B-Instruct-2507 demonstrates moderate speed performance, ranking in the 34th percentile across various benchmarks. It generally offers cost-effective solutions, placing in the 61st percentile for pricing. A standout feature is its exceptional reliability, boasting a 98% success rate, indicating consistent and usable responses. In terms of specific benchmarks, the model exhibits strong capabilities in Coding (85.9% accuracy, 68th percentile) and Reasoning (75.0% accuracy, 73rd percentile). It achieves near-perfect accuracy in General Knowledge (99.5%, 84th percentile) and a perfect 100% in Ethics, notably being the most accurate model at its price point and among models of similar speed. However, its performance in Instruction Following (39.7% accuracy, 39th percentile) and Email Classification (94.0% accuracy, 32nd percentile) is less competitive, suggesting areas for improvement in precision for complex multi-step instructions and classification tasks despite high raw accuracy in the latter. Overall, the model excels in knowledge-intensive and logical reasoning tasks, showcasing its strength in complex problem-solving and ethical considerations.

Model Pricing

Current Pricing

Feature Price (per 1M tokens)
Prompt $0.078
Completion $0.312

Price History

Available Endpoints
Provider Endpoint Name Context Length Pricing (Input) Pricing (Output)
Parasail
Parasail | qwen/qwen3-235b-a22b-07-25 262K $0.078 / 1M tokens $0.312 / 1M tokens
DeepInfra
DeepInfra | qwen/qwen3-235b-a22b-07-25 262K $0.13 / 1M tokens $0.6 / 1M tokens
Targon
Targon | qwen/qwen3-235b-a22b-07-25 262K $0.078 / 1M tokens $0.312 / 1M tokens
Parasail
Parasail | qwen/qwen3-235b-a22b-07-25 262K $0.15 / 1M tokens $0.85 / 1M tokens
Fireworks
Fireworks | qwen/qwen3-235b-a22b-07-25 262K $0.22 / 1M tokens $0.88 / 1M tokens
Targon
Targon | qwen/qwen3-235b-a22b-07-25 262K $0.12 / 1M tokens $0.59 / 1M tokens
Alibaba
Alibaba | qwen/qwen3-235b-a22b-07-25 131K $0.7 / 1M tokens $2.8 / 1M tokens
Together
Together | qwen/qwen3-235b-a22b-07-25 262K $0.2 / 1M tokens $0.6 / 1M tokens
Novita
Novita | qwen/qwen3-235b-a22b-07-25 262K $0.078 / 1M tokens $0.312 / 1M tokens
GMICloud
GMICloud | qwen/qwen3-235b-a22b-07-25 131K $0.17 / 1M tokens $1.09 / 1M tokens
Novita
Novita | qwen/qwen3-235b-a22b-07-25 262K $0.15 / 1M tokens $0.8 / 1M tokens
Cerebras
Cerebras | qwen/qwen3-235b-a22b-07-25 131K $0.6 / 1M tokens $1.2 / 1M tokens
Chutes
Chutes | qwen/qwen3-235b-a22b-07-25 262K $0.078 / 1M tokens $0.312 / 1M tokens
Nebius
Nebius | qwen/qwen3-235b-a22b-07-25 262K $0.2 / 1M tokens $0.6 / 1M tokens
BaseTen
BaseTen | qwen/qwen3-235b-a22b-07-25 262K $0.22 / 1M tokens $0.8 / 1M tokens
AtlasCloud
AtlasCloud | qwen/qwen3-235b-a22b-07-25 262K $0.35 / 1M tokens $1.2 / 1M tokens
Benchmark Results
Benchmark Category Reasoning Free Executions Accuracy Cost Duration
Other Models by qwen