上下文窗口效率探索器

分析300个AI模型的评分/上下文令牌比率，找出最充分利用上下文窗口、输出容量和成本的模型。

上下文窗口 vs 评分

LMMarketCap.com

效率概览

所有分析模型的关键效率指标。

最高效 (128K+)

GPT-5.3 Chat

707.0 score/MToken

最佳输出效率

Gemma 2 27B

37.8 score/1K output

最佳成本效率

Ling-2.6-flash

2000.0 score/$

平均总体效率

4.4%

所有模型标准化

效率排名

按每百万上下文令牌评分排名的前50个模型。

#	模型	提供商	评分	上下文	输出	评分/百万令牌	输出效率	成本效率	层级
1	Gemma 2 27BGoogle	Google	77	8K	2K	9448.2	37.8	119.1	Small
2	GPT-4OpenAI	OpenAI	65	8K	4K	7911.1	15.8	1.4	Small
3	R1 Distill Llama 70BDeepSeek	DeepSeek	41	8K	8K	4968.3	5.0	50.9	Small
4	Phi 4Microsoft	Microsoft	60	16K	16K	3674.3	3.7	573.3	Small
5	Reka Edgerekaai	rekaai	40	16K	16K	2441.4	2.4	400.0	Small
6	Perceptron Mk1perceptron	perceptron	40	33K	8K	1220.7	4.9	48.5	Medium
7	MiniMax M2-herMiniMax	MiniMax	69	66K	2K	1054.4	33.7	92.1	Medium
8	Mixtral 8x22B InstructMistral AI	Mistral AI	63	66K	--	967.4	--	15.8	Medium
9	GLM 4.5VZhipu AI	Zhipu AI	63	66K	16K	953.7	3.8	52.1	Medium
10	Olmo 3 32B ThinkAllen AI	Allen AI	55	66K	66K	837.7	0.8	168.9	Medium
11	Qwen3 30B A3B Thinking 2507Alibaba	Alibaba	64	82K	33K	782.5	2.0	49.3	Medium
12	GPT-5.3 ChatOpenAI	OpenAI	91	128K	16K	707.0	5.5	11.5	Large
13	GPT-5.2 ChatOpenAI	OpenAI	91	128K	16K	707.0	5.5	11.5	Large
14	GLM 4.5Zhipu AI	Zhipu AI	75	131K	98K	573.0	0.8	53.6	Large
15	GPT-4o (batch)OpenAI	OpenAI	72	128K	16K	562.5	4.4	23.0	Large
16	GPT-4o (2024-11-20)OpenAI	OpenAI	71	128K	16K	556.3	4.3	11.4	Large
17	GPT-4o (2024-08-06)OpenAI	OpenAI	71	128K	16K	556.3	4.3	11.4	Large
18	GPT-4oOpenAI	OpenAI	71	128K	16K	556.3	4.3	11.4	Large
19	GPT-4o (2024-05-13)OpenAI	OpenAI	71	128K	4K	556.3	17.4	7.1	Large
20	GPT-4o-miniOpenAI	OpenAI	69	128K	16K	541.4	4.2	184.8	Large
21	GLM 4.5 AirZhipu AI	Zhipu AI	71	131K	98K	539.4	0.7	144.3	Large
22	Qwen3 VL 235B A22B ThinkingAlibaba	Alibaba	69	131K	33K	529.5	2.1	31.5	Large
23	GPT-4 TurboOpenAI	OpenAI	67	128K	4K	521.1	16.3	3.3	Large
24	Mistral LargeMistral AI	Mistral AI	66	128K	--	514.8	--	16.5	Large
25	Llama 3.3 70B InstructMeta	Meta	67	131K	16K	509.6	4.1	318.1	Large
26	GPT-4 Turbo PreviewOpenAI	OpenAI	65	128K	4K	506.2	15.8	3.2	Large
27	Mistral Large 2407Mistral AI	Mistral AI	66	131K	--	502.8	--	16.5	Large
28	GLM 4.6VZhipu AI	Zhipu AI	66	131K	33K	499.7	2.0	109.2	Large
29	Llama 3.1 70B InstructMeta	Meta	65	131K	16K	498.2	4.0	163.3	Large
30	DeepSeek V3.2DeepSeek	DeepSeek	81	164K	66K	496.2	1.2	243.0	Large
31	Qwen3 30B A3BAlibaba	Alibaba	64	131K	16K	489.0	3.9	206.8	Large
32	GPT-4o-mini (batch)OpenAI	OpenAI	63	128K	16K	488.3	3.8	333.3	Large
33	R1 0528DeepSeek	DeepSeek	79	164K	33K	484.6	2.4	59.9	Large
34	GPT-4 Turbo (batch)OpenAI	OpenAI	62	128K	4K	480.5	15.0	6.2	Large
35	Mercury 2Inception	Inception	61	128K	50K	476.6	1.2	122.0	Large
36	Qwen3 8BAlibaba	Alibaba	61	131K	8K	465.4	7.4	213.3	Large
37	R1DeepSeek	DeepSeek	74	164K	16K	452.9	4.6	46.4	Large
38	GPT-4o-mini (2024-07-18)OpenAI	OpenAI	57	128K	16K	441.4	3.4	150.7	Large
39	DeepSeek V3.2 ExpDeepSeek	DeepSeek	72	164K	66K	438.2	1.1	211.2	Large
40	DeepSeek V3.1DeepSeek	DeepSeek	72	164K	33K	438.2	2.2	119.7	Large
41	DeepSeek V3 0324DeepSeek	DeepSeek	72	164K	66K	438.2	1.1	103.3	Large
42	gpt-oss-20bOpenAI	OpenAI	57	131K	131K	437.9	0.4	717.5	Large
43	gpt-oss-20b (free)OpenAI	OpenAI	57	131K	33K	437.9	1.8	Free	Large
44	o3 ProOpenAI	OpenAI	87	200K	100K	433.5	0.9	1.7	Large
45	o3 Pro (batch)OpenAI	OpenAI	87	200K	100K	433.5	0.9	3.5	Large
46	o3OpenAI	OpenAI	87	200K	100K	433.5	0.9	17.3	Large
47	o3 (batch)OpenAI	OpenAI	87	200K	100K	433.5	0.9	34.7	Large
48	Claude Opus 4.5Anthropic	Anthropic	85	200K	64K	425.5	1.3	5.7	Large
49	Claude Opus 4.5 (batch)Anthropic	Anthropic	85	200K	64K	425.5	1.3	11.3	Large
50	DeepSeek V3DeepSeek	DeepSeek	70	164K	16K	424.2	4.3	108.1	Large

层级分析

不同上下文窗口层级的效率分析。

Small5 models

平均评分57

评分/百万令牌5688.7

最佳

Gemma 2 27B

最差

Reka Edge

Medium6 models

平均评分59

评分/百万令牌969.4

最佳

Perceptron Mk1

最差

Qwen3 30B A3B Thinking 2507

Large174 models

平均评分65

评分/百万令牌311.8

最佳

GPT-5.3 Chat

最差

Grok Latest

Mega115 models

平均评分73

评分/百万令牌73.1

最佳

GLM 5.2 (batch)

最差

OpenAI GPT Latest

边际递减分析

更大的上下文窗口是否与更高的评分相关？

层级	平均上下文	平均评分	平均效率
Small	11K	57	5688.7
Medium	63K	59	969.4
Large	236K	65	311.8
Mega	1.0M	73	73.1

输出令牌效率

按输出效率（每1K输出令牌评分）排名的前20个模型。16K+输出令牌的模型已高亮显示。

模型	提供商	评分	最大输出	输出效率
Gemma 2 27BGoogle	Google	77	2K	37.8
MiniMax M2-herMiniMax	MiniMax	69	2K	33.7
GPT-4o (2024-05-13)OpenAI	OpenAI	71	4K	17.4
GPT-4 TurboOpenAI	OpenAI	67	4K	16.3
GPT-4 Turbo PreviewOpenAI	OpenAI	65	4K	15.8
GPT-4OpenAI	OpenAI	65	4K	15.8
GPT-4 Turbo (batch)OpenAI	OpenAI	62	4K	15.0
Claude 3 HaikuAnthropic	Anthropic	51	4K	12.5
Command R (08-2024)Cohere	Cohere	49	4K	12.2
Command R+ (08-2024)Cohere	Cohere	49	4K	12.2
Qwen3 8BAlibaba	Alibaba	61	8K	7.4
Qwen3 235B A22BAlibaba	Alibaba	54	8K	6.6
Command ACohere	Cohere	51	8K	6.2
GPT-5.3 ChatOpenAI16K+	OpenAI	91	16K	5.5
GPT-5.2 ChatOpenAI16K+	OpenAI	91	16K	5.5
R1 Distill Llama 70BDeepSeek	DeepSeek	41	8K	5.0
Nemotron 3.5 Content Safety (free)NVIDIA	NVIDIA	40	8K	4.9
Perceptron Mk1perceptron	perceptron	40	8K	4.9
Palmyra X5Writer	Writer	40	8K	4.9
Falcon-H1-Arabic 34B InstructTII	TII	40	8K	4.9

关键洞察

从效率数据中自动生成的观察结果。

上下文最优点

Small models have the highest average efficiency at 5688.7 score/MToken across 5 models.

输出很重要

Models with 16K+ output tokens score 29% higher on average than models with smaller output limits.

紧凑型高性能模型

0 models achieve top-20 scores with under 128K context.

探索更多

深入了解上下文窗口、对比模型或探索其他维度。

全部探索器上下文窗口大上下文模型

Frequently Asked Questions

Efficiency is measured as the score-per-context-token ratio - how much ranking score a model achieves relative to its context window size. Models that score highly with smaller context windows are considered more efficient than those requiring massive context to achieve similar results.

Cost efficiency combines quality (composite score) with pricing. The most cost-efficient models achieve high benchmark scores while maintaining low per-token API costs. Free and budget-tier models that perform well are the most cost-efficient options.

Not necessarily. Our efficiency analysis shows diminishing returns beyond certain context sizes. Models with 128K tokens often score similarly to those with 1M+ tokens, meaning the extra context capacity adds cost without proportional quality gains for most use cases.