geekradar
Int
Explore GeekRadar

More to explore

Good gear All topics About us Say hello
GeekRadar / The comparison desk

COMPARE.

AI
Filters

Filter AI models

Brand
Benchmark coverage
Clear all filters

Best result per entry; configurations may differ.

Type
23 results Benchmark score
anthropic

Claude Opus 5

Models
Context window
1,000,000 tokens
Maximum output
128,000 tokens
Open weights
No
Maximum input
1,000,000 tokens*
Reported score · version unreported · intelligence#1* · 61*Platform unreported
ECI 164.5*General capability index
openai

GPT-6 Astra

Models
Context window
1,050,000 tokens
Maximum output
128,000 tokens
Open weights
No
Maximum input
922,000 tokens*
Reported score · version unreported · intelligence#1* · 61*Platform unreported
Arena 1441.2Human preference · 15,038 votes
anthropic

Claude Fable 5

Models
Context window
1,000,000 tokens
Maximum output
128,000 tokens
Open weights
No
Maximum input
1,000,000 tokens*
Reported score · version unreported · intelligence#3* · 60*Platform unreported
ECI 163.0*General capability index
openai

GPT-5.6 Sol

Models
Context window
1,050,000 tokens
Maximum output
128,000 tokens
Open weights
No
Maximum input
922,000 tokens*
Reported score · version unreported · intelligence#4* · 59*Platform unreported
ECI 162.6*General capability index
moonshotai

Kimi K3

Models
Context window
1,048,576 tokens
Maximum output
131,072 tokens
Open weights
Yes
Maximum input
1,000,000 tokens*
Reported score · version unreported · intelligence#5* · 57*Platform unreported
ECI 158.3*General capability index
anthropic

Claude Opus 4.8

Models
Context window
1,000,000 tokens
Maximum output
128,000 tokens
Open weights
No
Maximum input
1,000,000 tokens*
Reported score · version unreported · intelligence#6* · 55.70*Platform unreported
ECI 159.6*General capability index
openai

GPT-5.5

Models
Context window
1,050,000 tokens
Maximum output
128,000 tokens
Open weights
No
Maximum input
1,050,000 tokens*
Reported score · version unreported · intelligence#7* · 54.80*Platform unreported
ECI 159.4*General capability index
Unspecified

GLM 5.2

Models*
Open weights
Yes*
Declared context window
1M*
Model type
LLM (coding/agentic, MoE)*
Declared modalities
Text → Text*
Reported score · version unreported · intelligence#8* · 51.10*Platform unreported
Radar · unrated0/4 core tests · not enough evidence
google

Gemini 3.5 Flash

Models
Context window
1,048,576 tokens
Maximum output
65,536 tokens
Open weights
No
Maximum input
1,048,576 tokens*
Reported score · version unreported · intelligence#9* · 50.20*Platform unreported
ECI 155.0*General capability index
anthropic

Claude Sonnet 4.6

Models
Context window
1,000,000 tokens
Maximum output
64,000 tokens
Open weights
No
Maximum input
1,000,000 tokens*
Reported score · version unreported · intelligence#10* · 47*Platform unreported
ECI 152.9*General capability index
google

Gemini 3.1 Pro

Models
Open weights
No*
Declared context window
1.05M*
Model type
LLM (multimodal)*
Declared modalities
Text, Image, Audio, Video → Text*
Reported score · version unreported · intelligence#11* · 46.50*Platform unreported
ECI 155.2*General capability index
deepseek

DeepSeek V4 Pro

Models
Context window
1,000,000 tokens
Maximum output
384,000 tokens
Open weights
Yes
Maximum input
1,000,000 tokens*
Reported score · version unreported · intelligence#12* · 44.30*Platform unreported
ECI 149.1*General capability index
moonshotai

Kimi K2.7 Code

Models
Context window
262,144 tokens
Maximum output
262,144 tokens
Open weights
Yes
Reasoning
Yes
Reported score · version unreported · intelligence#13* · 42*Platform unreported
ECI 150.2*General capability index
deepseek

DeepSeek V4 Flash

Models
Context window
1,000,000 tokens
Maximum output
384,000 tokens
Open weights
Yes
Maximum input
1,000,000 tokens*
Reported score · version unreported · intelligence#14* · 40.30*Platform unreported
ECI 146.2*General capability index
anthropic

Claude Haiku 4.5

Models
Context window
200,000 tokens
Maximum output
64,000 tokens
Open weights
No
Maximum input
200,000 tokens*
Reported score · version unreported · intelligence#15* · 37*Platform unreported
Arena 1396.6Human preference · 149,199 votes
deepseek

DeepSeek-R1

Models
Context window
128,000 tokens
Maximum output
32,768 tokens
Open weights
Yes
Maximum input
65,536 tokens*
Reported score · version unreported · intelligence#16* · 20.10*Platform unreported
ECI 138.8*General capability index
mistral

Mistral Large 3

Models
Context window
262,144 tokens
Maximum output
262,144 tokens
Open weights
Yes
Maximum input
262,144 tokens*
Reported score · version unreported · intelligence#17* · 15.90*Platform unreported
Arena 1428.3Human preference · 88,325 votes
meta

Llama 4 Maverick

Models
Open weights
Yes*
Declared context window
1M*
Model type
Multimodal (MoE)*
Declared modalities
Text, Image → Text*
Reported score · version unreported · intelligence#18* · 14.30*Platform unreported
ECI 132.3*General capability index
alibaba

Qwen3 235B-A22B

Models
Context window
131,072 tokens
Maximum output
16,384 tokens
Total parameters
235.09 B
Open weights
Yes
Reported score · version unreported · intelligence#19* · 13*Platform unreported
ECI 139.1*General capability index
alibaba

Qwen3 32B

Models
Context window
131,072 tokens
Maximum output
16,384 tokens
Total parameters
32.76 B
Open weights
Yes
Reported score · version unreported · intelligence#20* · 12*Platform unreported
Arena 1340.1Human preference · 3,926 votes
meta

Llama 4 Scout

Models
Open weights
Yes*
Declared context window
10M*
Model type
Multimodal (MoE)*
Declared modalities
Text, Image → Text*
Reported score · version unreported · intelligence#21* · 10*Platform unreported
ECI 129.6*General capability index
google

Gemma 3 27B

Models
Open weights
Yes*
Declared context window
128K*
Model type
LLM (multimodal)*
Declared modalities
Text, Image → Text*
Reported score · version unreported · intelligence#22* · 7.40*Platform unreported
ECI 129.9*General capability index
Unspecified

Phi-4

Models*
Open weights
Yes*
Declared context window
16K*
Model type
LLM (dense)*
Declared modalities
Text → Text*
Reported score · version unreported · intelligence#23* · 4.90*Platform unreported
Radar · unrated0/4 core tests · not enough evidence

* Provisional data