The best open LLMs for your use case:

1MiniMax M3.1 FlashMiniMax

MiniMax's natively multimodal 428B-A23B MoE built for fast, low-cost coding, agentic workflows, and computer use across a 1M-token context.

Speed:

Intelligence:

Price: (1M Tokens)

$0.15 / 0.60

Cached input: (1M Tokens)

$0.015

Context: (tokens)

1,048,576

Inputs:

ImageText

Benchmarks:

#3

HLE

General Knowledge

43.4
#3

DeepSWE

Coding Agents

73.7
#3

HLE

General Knowledge

43.4
#3

DeepSWE

Coding Agents

73.7
Try it out

Use case:

Personal Agents

Features:

Low Latency