Beta
Pricing · Hidden GemsTop-tier score · below-median price · low mindshare · computed live

Hidden Gems · Underpriced AI

Models that punch above their mindshare. Top-20 benchmarks, below-median price, low English- language coverage. Where the AI model market is mispricing quality.

Criterion 1
Top-20 benchmark
Average score ranks within top 30 of all priced models.
Criterion 2
Below-median price
Input ≤ $$0.400/M (median across all priced).
Criterion 3
Low mindshare
Excludes OpenAI, Anthropic, Google flagships · where hype already priced in.
10 gems found
Value 827.5
Score
66.2
Price
$0.080
Context
131K

Benchmark rank #20 of 145. Open weights · self-hostable. Provider Alibaba Qwen receives a fraction of the English-language AI coverage of frontier labs · the gap between attention and capability is where this model is mispriced.

Value 814.4
Score
73.3
Price
$0.090
Context
262K

Benchmark rank #6 of 145. Open weights · self-hostable. Provider xiaomi receives a fraction of the English-language AI coverage of frontier labs · the gap between attention and capability is where this model is mispriced.

Value 769.0
Score
76.9
Price
$0.100
Context
262K

Benchmark rank #4 of 145. Open weights · self-hostable. Provider stepfun receives a fraction of the English-language AI coverage of frontier labs · the gap between attention and capability is where this model is mispriced.

Value 631.8
Score
61.6
Price
$0.098
Context
131K

Benchmark rank #26 of 145. Open weights · self-hostable. Provider Alibaba Qwen receives a fraction of the English-language AI coverage of frontier labs · the gap between attention and capability is where this model is mispriced.

minimax logo
MiniMax M2
minimax
Value 272.5
Score
69.5
Price
$0.255
Context
197K

Benchmark rank #13 of 145. Open weights · self-hostable. Provider minimax receives a fraction of the English-language AI coverage of frontier labs · the gap between attention and capability is where this model is mispriced.

Alibaba Qwen logo
Qwen3.6 Plus
Alibaba Qwen
Value 218.2
Score
70.9
Price
$0.325
Context
1.0M

Benchmark rank #9 of 145. Open weights · self-hostable. Provider Alibaba Qwen receives a fraction of the English-language AI coverage of frontier labs · the gap between attention and capability is where this model is mispriced.

Value 216.0
Score
64.8
Price
$0.300
Context
131K

Benchmark rank #22 of 145. Provider xAI receives a fraction of the English-language AI coverage of frontier labs · the gap between attention and capability is where this model is mispriced.

minimax logo
Value 211.7
Score
63.5
Price
$0.300
Context
197K

Benchmark rank #24 of 145. Open weights · self-hostable. Provider minimax receives a fraction of the English-language AI coverage of frontier labs · the gap between attention and capability is where this model is mispriced.

Alibaba Qwen logo
Qwen3.5 397B A17B
Alibaba Qwen
Value 201.0
Score
78.4
Price
$0.390
Context
262K

Benchmark rank #2 of 145. Open weights · self-hostable. Provider Alibaba Qwen receives a fraction of the English-language AI coverage of frontier labs · the gap between attention and capability is where this model is mispriced.

Value 195.5
Score
78.2
Price
$0.400
Context
164K

Benchmark rank #3 of 145. Open weights · self-hostable. Provider DeepSeek receives a fraction of the English-language AI coverage of frontier labs · the gap between attention and capability is where this model is mispriced.

A model with top-20 benchmark rank, below-median price, and low mindshare. The ranking combines score vs score-rank delta, price vs price-rank delta, and provider attention · when a model has high performance but few people talk about it, that gap is opportunity.