Value ranking
Best value on ARC-AGI 2
As of October 25, 2026, the best value on ARC-AGI 2 is DeepSeek V4.1 Flash, at 3837.50 normalized points per dollar of input price ($0.02 per million tokens, score 76.8).
Second-generation ARC challenge testing fluid reasoning over abstract visual puzzles. Resists training-data memorisation by construction: each puzzle is novel and solutions require multi-step pattern induction. Frontier models are only just starting to score above chance on the harder tier.
“Value” is normalized benchmark score (0–100 for this leaderboard cohort) divided by input price per million tokens. Higher means more capability per dollar on this axis only — always sanity-check latency, context length, and your real workload.
- 1DeepSeek V4.1 FlashDeepSeek3837.5076.8 / $0.02/M
- 2DeepSeek V4 Flash 0731DeepSeek1531.2864.6 / $0.04/M
- 3GPT-6 LunaOpenAI1248.6062.4 / $0.05/M
- 4Claude Haiku 5.5Anthropic644.7064.5 / $0.10/M
- 5GPT-5.6 LunaOpenAI626.7062.7 / $0.10/M
- 6DeepSeek V4 Flash 0423DeepSeek531.7648.4 / $0.09/M
- 7GLM 5.3 FlashZhipu AI (GLM)462.0069.3 / $0.15/M
- 8DeepSeek V4 Pro 0423DeepSeek283.5759.2 / $0.21/M
- 9Gemini 3.8 FlashGoogle250.2993.9 / $0.38/M
- 10Gemini 3.7 FlashGoogle237.4489.0 / $0.38/M
- 11Gemini 3.6 FlashGoogle169.6063.6 / $0.38/M
- 12Gemini 3 FlashGoogle141.5235.4 / $0.25/M
- 13GPT-5 nanoOpenAI110.002.8 / $0.03/M
- 14Qwen3.8 27BAlibaba (Qwen)104.9244.6 / $0.43/M
- 15Gemini 3.5 FlashGoogle101.1675.9 / $0.75/M
- 16GPT-6.1 SolOpenAI99.1299.1 / $1.00/M
- 17DeepSeek V4 Pro 0813DeepSeek97.6864.5 / $0.66/M
- 18Gemini 4 ArgonGoogle96.0596.0 / $1.00/M
- 19GPT-6 SolOpenAI94.3094.3 / $1.00/M
- 20Inkling SmallThinking Machines93.8942.3 / $0.45/M
- 21GPT-5.6 TerraOpenAI88.3288.3 / $1.00/M
- 22Gemini 3.1 ProGoogle81.1681.2 / $1.00/M
- 23Gemini 3.5 Flash LiteGoogle72.1310.8 / $0.15/M
- 24GPT-5.2OpenAI63.6555.7 / $0.88/M
- 25GPT-5.4OpenAI62.2777.8 / $1.25/M
AI Model Analyzer does not recommend specific vendors; rankings are derived from public data only.