COMPARE.

Unspecified

mistral.magistral-small-2509

Models^*

Maximum output

40,000 tokens^*

Maximum input

128,000 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

mistral.ministral-3-14b-instruct

Models^*

Maximum output

8,000 tokens^*

Maximum input

128,000 tokens^*

Tool calling

Yes^*

Structured output

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

mistral.ministral-3-3b-instruct

Models^*

Maximum output

8,000 tokens^*

Maximum input

128,000 tokens^*

Tool calling

Yes^*

Structured output

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

mistral.ministral-3-8b-instruct

Models^*

Maximum output

8,000 tokens^*

Maximum input

128,000 tokens^*

Tool calling

Yes^*

Structured output

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

mistral.mistral-large-3-675b-instruct

Models^*

Maximum output

32,000 tokens^*

Maximum input

256,000 tokens^*

Tool calling

Yes^*

Structured output

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

moonshot.kimi-k2-thinking

Models^*

Maximum output

16,000 tokens^*

Maximum input

256,000 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

nvidia

nvidia.nemotron-nano-9b-v2

Models^*

Maximum output

8,000 tokens^*

Maximum input

128,000 tokens^*

Tool calling

Yes^*

Structured output

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

qwen.qwen3-coder-next

Models^*

Maximum output

16,000 tokens^*

Maximum input

256,000 tokens^*

Tool calling

Yes^*

Structured output

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

sail/deepseek-ai/DeepSeek-V4-Flash-0731

Models^*

Maximum output

1,048,576 tokens^*

Maximum input

1,048,576 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

sail/deepseek-ai/DeepSeek-V4-Pro-0813

Models^*

Maximum output

1,048,576 tokens^*

Maximum input

1,048,576 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

sail/deepseek-ai/DeepSeek-V4.1-Flash

Models^*

Maximum output

1,048,576 tokens^*

Maximum input

1,048,576 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

sail/google/gemma-4-12B-it

Models^*

Maximum output

16,384 tokens^*

Maximum input

16,384 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

sail/moonshotai/Kimi-K3

Models^*

Maximum output

1,048,576 tokens^*

Maximum input

1,048,576 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

sail/zai-org/GLM-5.3

Models^*

Maximum output

1,048,576 tokens^*

Maximum input

1,048,576 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

sail/zai-org/GLM-5.3-Flash

Models^*

Maximum output

1,048,576 tokens^*

Maximum input

1,048,576 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

tencent/glm-5.2

Models^*

Maximum output

128,000 tokens^*

Maximum input

1,000,000 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

tencent/glm-5.3

Models^*

Maximum output

128,000 tokens^*

Maximum input

1,000,000 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

tencent/kimi-k2.7-code

Models^*

Maximum output

256,000 tokens^*

Maximum input

256,000 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

tencent/kimi-k2.7-code-highspeed

Models^*

Maximum output

256,000 tokens^*

Maximum input

256,000 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

tencent/kimi-k2.8-preview

Models^*

Maximum output

1,000,000 tokens^*

Maximum input

1,000,000 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

tencent/kimi-k3

Models^*

Maximum output

1,000,000 tokens^*

Maximum input

1,000,000 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

vercel_ai_gateway/anthropic/claude-opus-4.5

Models^*

Maximum output

64,000 tokens^*

Maximum input

200,000 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

vercel_ai_gateway/google/gemini-2.5-pro

Models^*

Maximum output

65,536 tokens^*

Maximum input

1,048,576 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence

Unspecified

xai/grok-4.20-beta

Models^*

Maximum output

1,000,000 tokens^*

Maximum input

1,000,000 tokens^*

Reasoning

Yes^*

Tool calling

Yes^*

Radar · unrated0/4 core tests · not enough evidence