Best AI models for coding

The top AI models for real-world coding, ranked by SWE-Bench Verified — which measures how often a model resolves actual GitHub issues. Every one is available on anyAInow from a single balance; pay only for what you use, no subscription.

#ModelSWE-Bench Verified
1Claude Opus 4.8
AnthropicFlagship·from $40 / 1M
88.6%Try →
2Claude Opus 4.7
AnthropicFlagship·from $40 / 1M
87.6%Try →
3Gemini 3.1 Pro
GoogleFlagship·from $20 / 1M
80.6%Try →
4MiniMax M3
MiniMaxOpen·from $10 / 1M
80.5%Try →
5Kimi K2
Moonshot (Kimi)Balanced·from $10 / 1M
80.2%Try →
6Claude Sonnet 4.6
AnthropicBalanced·from $30 / 1M
79.6%Try →
7MiMo V2 Pro
Xiaomi (MiMo)Open·from $10 / 1M
78.9%Try →
8Qwen 3.6 Plus
QwenOpen·from $10 / 1M
78.8%Try →
9Qwen 3.7 Plus
QwenOpen·from $10 / 1M
77.7%Try →
10Mistral Medium
MistralBalanced·from $20 / 1M
77.6%Try →
11Claude Haiku 4.5
AnthropicFast·from $10 / 1M
73.3%Try →
12Gemini 2.5 Pro
GoogleFlagship·from $20 / 1M
63.2%Try →
13DeepSeek R1
DeepSeekReasoning·from $10 / 1M
44.6%Try →

Benchmarks via llm-stats.com, as of 28 Jun 2026. Figures are publisher-reported where not independently verified. Prices shown are the anyAInow pay-as-you-go floor (per 1M tokens); free-tier models deduct nothing.

More rankings