Leaderboard

Model rankings / 2026

Independent benchmark scores, divided by what this market actually charges.

Ability

Performance

Price

Per dollar

This market

Intelligence per dollar

The intelligence index divided by what a million tokens cost here — the ranking only this market can publish.

glm-5.3-flash1760% below list

gpt-5.6-luna33395% below list

gpt-5.6-terra15095% below list

Top 10

The strongest model each vendor brings to this ranking. Every axis is scaled to the best on the board, so the outer ring is always the leader.

No model has every axis measured yet.

Every model twice: where the vendor prices it, and where it trades here.

OpenAIAnthropicGoogleMoonshotxAI

Vendor list pricePrice on AsaleValue frontier

#	Model	Intelligence per dollar	Intelligence	Price / 1M	Saving
1	gpt-5.6-lunaGPT-5.6 Luna (max)	333	37.5	$0.113	95%
2	glm-5.3-flashGLM-5.3-Flash	176	41.9	$0.238	—
3	gpt-5.6-terraGPT-5.6 Terra (max)	150	42.3	$0.281	95%
4	nemotron-3.5-lightningNemotron 3.5 Lightning	124	13.6	$0.11	—
5	qwen3.5-9bQwen3.5 9B (Reasoning)	122	13.7	$0.113	—
6	glm-4.7-flashGLM-4.7-Flash (Reasoning)	103	14.9	$0.145	—
7	gpt-5-nanoGPT-5 nano (high)	94.5	13.0	$0.138	—
8	gemma-4-26b-a4b-itGemma 4 26B A4B (Reasoning)	94.1	16.7	$0.178	—
9	gpt-5.6-solGPT-5.6 Sol (max)	83.7	47.1	$0.563	95%
10	qwen3-coder-30b-a3b-instructQwen3 Coder 30B A3B Instruct	80.0	9.6	$0.12	—
11	gemma-3-4b-itGemma 3 4B Instruct	76.8	4.8	$0.063	—
12	deepseek-v4.1-flashDeepSeek V4.1 Flash (Reasoning, Max Effort)	75.2	39.5	$0.525	—
13	gemma-4-31b-itGemma 4 31B (Reasoning)	75.1	15.4	$0.205	—
14	gpt-5.5GPT-5.5 (xhigh)	68.6	38.6	$0.563	95%
15	gemma-3n-e4b-itGemma 3n E4B Instruct	64.0	4.8	$0.075	—
16	qwen3.5-35b-a3bQwen3.5 35B A3B (Reasoning)	54.4	19.3	$0.355	—
17	deepseek-v3.2DeepSeek V3.2 (Non-reasoning)	53.0	16.0	$0.302	—
18	qwen3.6-35b-a3bQwen3.6 35B A3B (Reasoning)	53.0	18.8	$0.355	—
19	gpt-6-astraGPT-6 Astra (max)	52.8	52.8	$1	95%
20	gemma-3-12b-itGemma 3 12B Instruct	50.7	3.8	$0.075	—
21	qwen3-vl-32b-instructQwen3 VL 32B Instruct	46.2	8.4	$0.182	—
22	qwen3.7-plusQwen3.7 Plus	46.1	25.8	$0.56	—
23	gpt-5.4-nanoGPT-5.4 nano (xhigh)	45.8	21.2	$0.463	—
24	llama-4-scoutLlama 4 Scout	43.3	6.5	$0.15	—
25	qwen3.5-27bQwen3.5 27B (Reasoning)	42.7	22.9	$0.536	—

One entry per model: catalog ids naming the same model collapse to whichever does best on the ranking shown.

Scores last pulled Sep 16, 2026.