Aya Expanse 32B: Intelligence, Performance & Price Analysis

Name: Intelligence Index by Aya Endpoint
Creator: Artificial Analysis

Analysis of Cohere's Aya Expanse 32B and comparison to other AI models across key metrics including quality, price, performance (tokens per second & time to first token), context window & more. Click on any model to compare API providers for that model. For more details including relating to our methodology, see our FAQs.

For analysis of API providers see

Comparison Summary

Intelligence:

Aya Expanse 32B is of lower quality compared to average, with a MMLU score of 0.377 and a Intelligence Index across evaluations of 20.

Price:

Aya Expanse 32B is cheaper compared to average with a price of $0.75 per 1M Tokens (blended 3:1).
Aya Expanse 32B Input token price: $0.50, Output token price: $1.50 per 1M Tokens.

Speed:

Aya Expanse 32B is faster compared to average, with a output speed of 119.6 tokens per second.

Latency:

Aya Expanse 32B has a lower latency compared to average, taking 0.15s to receive the first token (TTFT).

Context Window:

Aya Expanse 32B has a smaller context windows than average, with a context window of 130k tokens.

Highlights

Intelligence

Artificial Analysis Intelligence Index; Higher is better

Speed

Output Tokens per Second; Higher is better

Price

USD per 1M Tokens; Lower is better

Navigation

Intelligence

Context Window

Pricing

Performance Summary

Speed

Latency

Total Response Time

Comparisons

Aya Expanse 32B Model Details

Creator
Context Window	128k tokens (~192 A4 pages of size 12 Arial font)
Release Date	October, 2024
Parameters	32B
Image Input Support	No
Open Source (Weights)	Yes
License	CC-BY-NC 4.0 License with Acceptable Use Addendum
License Supports Commercial Use Without Restrictions	No

Intelligence

Back to Navigation

Artificial Analysis Intelligence Index

+ Add model from specific provider

Intelligence Index incorporates 7 evaluations spanning reasoning, knowledge, math & coding

Estimate (independent evaluation forthcoming)

Artificial Analysis Intelligence Index: Combination metric covering multiple dimensions of intelligence - the simplest way to compare how smart models are. Version 2 was released in Feb '25 and includes: MMLU-Pro, GPQA Diamond, Humanity's Last Exam, LiveCodeBench, SciCode, AIME, MATH-500. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Artificial Analysis Intelligence Index by Model Type

+ Add model from specific provider

Intelligence Index incorporates 7 evaluations spanning reasoning, knowledge, math & coding

Estimate (independent evaluation forthcoming)

Reasoning Model

Non-Reasoning Model

Artificial Analysis Intelligence Index by Open Weights vs Proprietary

+ Add model from specific provider

Intelligence Index incorporates 7 evaluations spanning reasoning, knowledge, math & coding

Estimate (independent evaluation forthcoming)

Proprietary

Open Weights

Open Weights (Commercial Use Restricted)

Open Weights: Indicates whether the model weights are available. Models are labelled as 'Commercial Use Restricted' if the weights are available but commercial use is limited (typically requires obtaining a paid license).

Artificial Analysis Coding Index

+ Add model from specific provider

Represents the average of coding benchmarks in the Artificial Analysis Intelligence Index (LiveCodeBench & SciCode)

Artificial Analysis Coding Index: Represents the average of coding evaluations in the Artificial Analysis Intelligence Index. Currently includes: LiveCodeBench, SciCode. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Artificial Analysis Math Index

+ Add model from specific provider

Represents the average of math benchmarks in the Artificial Analysis Intelligence Index (AIME 2024 & Math-500)

Artificial Analysis Math Index: Represents the average of math evaluations in the Artificial Analysis Intelligence Index. Currently includes: AIME, MATH-500. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Intelligence Evaluations

+ Add model from specific provider

Intelligence evaluations measured independently by Artificial Analysis; Higher is better

Results claimed by AI Lab (not yet independently verified)

MMLU-Pro (Reasoning & Knowledge)

GPQA Diamond (Scientific Reasoning)

Humanity's Last Exam (Reasoning & Knowledge)

LiveCodeBench (Coding)

SciCode (Coding)

HumanEval (Coding)

MATH-500 (Quantitative Reasoning)

AIME 2024 (Competition Math)

Multilingual Index (Artificial Analysis)

While model intelligence generally translates across use cases, specific evaluations may be more relevant for certain use cases.

Intelligence vs. Price

+ Add model from specific provider

Artificial Analysis Intelligence Index (Version 2, released Feb '25); Price: USD per 1M Tokens

Most attractive quadrant

GPT-4o (Nov '24)

GPT-4o mini

o3-mini (high)

Llama 3.3 70B

Llama 3.1 8B

Gemini 2.0 Flash

Claude 3.7 Sonnet

Claude 3.7 Sonnet Thinking

Mistral Large 2 (Nov '24)

Mistral Small 3.1

DeepSeek R1

DeepSeek V3 (Mar' 25)

Nova Pro

Nova Micro

Command A

Aya Expanse 32B

QwQ-32B

DeepSeek V3

While higher intelligence models are typically more expensive, they do not all follow the same price-quality curve.

Price: Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Intelligence vs. Output Speed

+ Add model from specific provider

Artificial Analysis Intelligence Index (Version 2, released Feb '25); Output Speed: Output Tokens per Second

Most attractive quadrant

GPT-4o (Nov '24)

GPT-4o mini

Llama 3.3 70B

Llama 3.1 8B

Gemini 2.0 Flash

Gemma 3 27B

Claude 3.7 Sonnet

Claude 3.7 Sonnet Thinking

Mistral Large 2 (Nov '24)

Mistral Small 3.1

DeepSeek R1

DeepSeek V3 (Mar' 25)

Nova Pro

Nova Micro

Command A

Aya Expanse 32B

QwQ-32B

DeepSeek V3

Output Speed: Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Intelligence vs. Total Response Time

+ Add model from specific provider

Artificial Analysis Intelligence Index (Version 2, released Feb '25); End-to-End Seconds to Output 100 Tokens; Lower is better

Most attractive quadrant

GPT-4o (Nov '24)

GPT-4o mini

Llama 3.3 70B

Llama 3.1 8B

Gemini 2.0 Flash

Gemma 3 27B

Claude 3.7 Sonnet

Claude 3.7 Sonnet Thinking

Mistral Large 2 (Nov '24)

Mistral Small 3.1

DeepSeek R1

DeepSeek V3 (Mar' 25)

Nova Pro

Nova Micro

Command A

Aya Expanse 32B

QwQ-32B

DeepSeek V3

Total Response Time: Time to receive a 100 token response. Calculated based on Latency (time to receive first token) and Output Speed (output tokens per second).

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Context Window

Back to Navigation

Context Window

+ Add model from specific provider

Context Window: Tokens Limit; Higher is better

Larger context windows are relevant to RAG (Retrieval Augmented Generation) LLM workflows which typically involve reasoning and information retrieval of large amounts of data.

Context window: Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

Intelligence vs. Context Window

+ Add model from specific provider

Artificial Analysis Intelligence Index (Version 2, released Feb '25); Context Window: Tokens Limit

Most attractive quadrant

GPT-4o (Nov '24)

GPT-4o mini

o3-mini (high)

Llama 3.3 70B

Llama 3.1 8B

Gemini 2.0 Flash

Gemma 3 27B

Claude 3.7 Sonnet

Claude 3.7 Sonnet Thinking

Mistral Large 2 (Nov '24)

Mistral Small 3.1

DeepSeek R1

DeepSeek V3 (Mar' 25)

Nova Pro

Nova Micro

Command A

Aya Expanse 32B

QwQ-32B

DeepSeek V3

Context window: Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

Pricing

Back to Navigation

Pricing: Input and Output Prices

+ Add model from specific provider

Price: USD per 1M Tokens

Input price

Output price

Input Price: Price per token included in the request/message sent to the API, represented as USD per million Tokens.

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Pricing Comparison of Aya Expanse 32B API Providers

Performance Summary

Back to Navigation

Output Speed vs. Price

+ Add model from specific provider

Output Speed: Output Tokens per Second; Price: USD per 1M Tokens

Most attractive quadrant

GPT-4o (Nov '24)

GPT-4o mini

Llama 3.3 70B

Llama 3.1 8B

Gemini 2.0 Flash

Claude 3.7 Sonnet

Claude 3.7 Sonnet Thinking

Mistral Large 2 (Nov '24)

Mistral Small 3.1

DeepSeek R1

DeepSeek V3 (Mar' 25)

Nova Pro

Nova Micro

Command A

Aya Expanse 32B

QwQ-32B

DeepSeek V3

Output Speed: Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Price: Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).

Latency vs. Output Speed

+ Add model from specific provider

Latency: Seconds to First Token Received; Output Speed: Output Tokens per Second

Most attractive quadrant

GPT-4o (Nov '24)

GPT-4o mini

Llama 3.3 70B

Llama 3.1 8B

Gemini 2.0 Flash

Gemma 3 27B

Gemini 2.5 Pro Experimental

Claude 3.7 Sonnet

Claude 3.7 Sonnet Thinking

Mistral Large 2 (Nov '24)

Mistral Small 3.1

DeepSeek R1

DeepSeek V3 (Mar' 25)

Nova Pro

Nova Micro

Command A

Aya Expanse 32B

QwQ-32B

DeepSeek V3

Output Speed: Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Latency: Time to first token of tokens received, in seconds, after API request sent. For models which do not support streaming, this represents time to receive the completion.

Price: Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).

Speed

Measured by Output Speed (tokens per second)

Back to Navigation

Output Speed

+ Add model from specific provider

Output Tokens per Second; Higher is better

Output Speed: Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Output Speed by Input Token Count (Context Length)

+ Add model from specific provider

Output Tokens per Second; Higher is better

100 input tokens

1k input tokens

10k input tokens

100k input tokens

Output Speed: Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Input Tokens Length: Length of tokens provided in the request. See Prompt Options above to see benchmarks of different input prompt lengths across other charts.

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Output Speed Variance

+ Add model from specific provider

Output Tokens per Second; Results by percentile; Higher is better

Median, Other points represent 5th, 25th, 75th, 95th Percentiles respectively

Output Speed: Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Boxplot: Shows variance of measurements

Output Speed, Over Time

+ Add model from specific provider

Output Tokens per Second; Higher is better

Output Speed: Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Over time measurement: Median measurement per day, based on 8 measurements each day at different times. Labels represent start of week's measurements.

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Latency

Measured by Time (seconds) to First Token

Back to Navigation

Latency

+ Add model from specific provider

Seconds to First Token Received; Lower is better

Latency: Time to first token of tokens received, in seconds, after API request sent. For models which do not support streaming, this represents time to receive the completion.

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Latency by Input Token Count (Context Length)

+ Add model from specific provider

Seconds to First Token Received; Lower is better

100 input tokens

1k input tokens

10k input tokens

100k input tokens

Input Tokens Length: Length of tokens provided in the request. See Prompt Options above to see benchmarks of different input prompt lengths across other charts.

Latency: Time to first token of tokens received, in seconds, after API request sent. For models which do not support streaming, this represents time to receive the completion.

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Latency Variance

+ Add model from specific provider

Seconds to First Token Received; Results by percentile; Lower is better

Median, Other points represent 5th, 25th, 75th, 95th Percentiles respectively

Latency: Time to first token of tokens received, in seconds, after API request sent. For models which do not support streaming, this represents time to receive the completion.

Boxplot: Shows variance of measurements

Total Response Time

Time to receive 100 tokens output, calculated from latency and output speed metrics

Back to Navigation

Total Response Time

+ Add model from specific provider

End-to-End Seconds to Output 100 Tokens; Lower is better

The speed difference between the fastest and slowest models is >3X. There is not always a correlation between parameter size and speed, or between price and speed.

Total Response Time: Time to receive a 100 token response. Calculated based on Latency (time to receive first token) and Output Speed (output tokens per second).

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Total Response Time by Input Token Count (Context Length)

+ Add model from specific provider

End-to-End Seconds to Output 100 Tokens; Lower is better

100 input tokens

1k input tokens

10k input tokens

100k input tokens

Input Tokens Length: Length of tokens provided in the request. See Prompt Options above to see benchmarks of different input prompt lengths across other charts.

Total Response Time: Time to receive a 100 token response. Calculated based on Latency (time to receive first token) and Output Speed (output tokens per second).

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Total Response Time Variance

+ Add model from specific provider

Total Response Time: End-to-End Seconds to Output 100 Tokens; Results by percentile; Lower is better

Median, Other points represent 5th, 25th, 75th, 95th Percentiles respectively

Total Response Time: Time to receive a 100 token response. Calculated based on Latency (time to receive first token) and Output Speed (output tokens per second).

Boxplot: Shows variance of measurements

Further details

Model Name	Creator	License	Context Window

o1	OpenAI	Proprietary	200k
o3-mini	OpenAI	Proprietary	200k
o1-preview	OpenAI	Proprietary	128k
o1-mini	OpenAI	Proprietary	128k
GPT-4o (Aug '24)	OpenAI	Proprietary	128k
GPT-4o (May '24)	OpenAI	Proprietary	128k
GPT-4o (Nov '24)	OpenAI	Proprietary	128k
GPT-4o mini	OpenAI	Proprietary	128k
o3	OpenAI	Proprietary	128k
GPT-4o (ChatGPT)	OpenAI	Proprietary	128k
o3-mini (high)	OpenAI	Proprietary	200k
GPT-4o Realtime (Dec '24)	OpenAI	Proprietary	128k
GPT-4.5 (Preview)	OpenAI	Proprietary	128k
GPT-4o mini Realtime (Dec '24)	OpenAI	Proprietary	128k
o1-pro	OpenAI	Proprietary	200k
GPT-4 Turbo	OpenAI	Proprietary	128k
GPT-4	OpenAI	Proprietary	8k

Llama 3.3 Instruct 70B	Meta	Open	128k
Llama 3.1 Instruct 405B	Meta	Open	128k
Llama 3.2 Instruct 90B (Vision)	Meta	Open	128k
Llama 3.1 Instruct 70B	Meta	Open	128k
Llama 3.2 Instruct 11B (Vision)	Meta	Open	128k
Llama 3.1 Instruct 8B	Meta	Open	128k
Llama 3.2 Instruct 3B	Meta	Open	128k
Llama 3.2 Instruct 1B	Meta	Open	128k
Llama 3 Instruct 70B	Meta	Open	8k
Llama 3 Instruct 8B	Meta	Open	8k
Llama 2 Chat 70B	Meta	Open	4k
Llama 2 Chat 13B	Meta	Open	4k
Llama 2 Chat 7B	Meta	Open	4k

Gemma 3 12B Instruct	Google	Open	128k
Gemma 3 4B Instruct	Google	Open	128k
Gemini 2.0 Pro Experimental (Feb '25)	Google	Proprietary	2m
Gemini 2.0 Flash (Feb '25)	Google	Proprietary	1m
Gemini 2.0 Flash (experimental)	Google	Proprietary	1m
Gemini 1.5 Pro (Sep '24)	Google	Proprietary	2m
Gemini 1.5 Flash (Sep '24)	Google	Proprietary	1m
Gemini 1.5 Pro (May '24)	Google	Proprietary	2m
Gemma 2 27B	Google	Open	8k
Gemma 2 9B	Google	Open	8k
Gemini 1.5 Flash-8B	Google	Proprietary	1m
Gemma 3 27B Instruct	Google	Open	128k
Gemini 2.0 Flash-Lite (Feb '25)	Google	Proprietary	1m
Gemini 2.0 Flash Thinking Experimental (Dec '24)	Google	Proprietary	2m
Gemini 2.5 Pro Experimental (Mar' 25)	Google	Proprietary	1m
Gemma 3 1B Instruct	Google	Open	32k
Gemini 2.0 Flash-Lite (Preview)	Google	Proprietary	1m
Gemini 1.0 Pro	Google	Proprietary	33k
Gemini 1.5 Flash (May '24)	Google	Proprietary	1m

Claude 3.5 Sonnet (Oct '24)	Anthropic	Proprietary	200k
Claude 3.5 Sonnet (June '24)	Anthropic	Proprietary	200k
Claude 3 Opus	Anthropic	Proprietary	200k
Claude 3.5 Haiku	Anthropic	Proprietary	200k
Claude 3 Haiku	Anthropic	Proprietary	200k
Claude 3.7 Sonnet (Standard)	Anthropic	Proprietary	200k
Claude 3.7 Sonnet (Extended Thinking)	Anthropic	Proprietary	200k
Claude 3 Sonnet	Anthropic	Proprietary	200k
Claude 2.1	Anthropic	Proprietary	200k
Claude 2.0	Anthropic	Proprietary	100k

Pixtral Large	Mistral	Open	128k
Mistral Large 2 (Nov '24)	Mistral	Open	128k
Mistral Large 2 (Jul '24)	Mistral	Open	128k
Mistral Small 3	Mistral	Open	32k
Mistral Small (Sep '24)	Mistral	Open	33k
Mixtral 8x22B Instruct	Mistral	Open	65k
Pixtral 12B (2409)	Mistral	Open	128k
Ministral 8B	Mistral	Open	128k
Mistral NeMo	Mistral	Open	128k
Ministral 3B	Mistral	Proprietary	128k
Mixtral 8x7B Instruct	Mistral	Open	33k
Codestral-Mamba	Mistral	Open	256k
Mistral Small 3.1	Mistral	Open	128k
Codestral (Jan '25)	Mistral	Proprietary	256k
Mistral Saba	Mistral	Proprietary	32k
Mistral Small (Feb '24)	Mistral	Proprietary	33k
Mistral Large (Feb '24)	Mistral	Proprietary	33k
Mistral 7B Instruct	Mistral	Open	8k
Codestral (May '24)	Mistral	Open	33k
Mistral Medium	Mistral	Proprietary	33k

DeepSeek R1	DeepSeek	Open	128k
DeepSeek R1 Distill Llama 70B	DeepSeek	Open	128k
DeepSeek-V2.5 (Dec '24)	DeepSeek	Open	128k
DeepSeek-Coder-V2	DeepSeek	Open	128k
DeepSeek LLM 67B Chat (V1)	DeepSeek	Open	4k
DeepSeek R1 Distill Qwen 32B	DeepSeek	Open	128k
DeepSeek V3 0324 (Mar' 25)	DeepSeek	Open	128k
DeepSeek Coder V2 Lite Instruct	DeepSeek	Open	128k
DeepSeek R1 Distill Qwen 1.5B	DeepSeek	Open	128k
DeepSeek R1 Distill Llama 8B	DeepSeek	Open	128k
DeepSeek R1 Distill Qwen 14B	DeepSeek	Open	128k
DeepSeek V3	DeepSeek	Open	128k
DeepSeek-V2.5	DeepSeek	Open	128k
DeepSeek-V2-Chat	DeepSeek	Open	128k

Sonar Reasoning Pro	Perplexity	Proprietary	127k
Sonar Reasoning	Perplexity	Proprietary	127k
Sonar	Perplexity	Proprietary	127k
Sonar Pro	Perplexity	Proprietary	200k

Grok Beta	xAI	Proprietary	128k
Grok 3 Reasoning Beta	xAI	Proprietary	1m
Grok 3 mini	xAI	Proprietary	1m
Grok 3 mini Reasoning	xAI	Proprietary	1m
Grok 3	xAI	Proprietary	1m
Grok 2 (Dec '24)	xAI	Proprietary	131k

Nova Pro	Amazon	Proprietary	300k
Nova Lite	Amazon	Proprietary	300k
Nova Micro	Amazon	Proprietary	130k

Phi-4	Microsoft Azure	Open	16k
Phi-3 Mini Instruct 3.8B	Microsoft Azure	Open	4k
Phi-4 Mini Instruct	Microsoft Azure	Open	128k
Phi-3 Medium Instruct 14B	Microsoft Azure	Open	128k
Phi-4 Multimodal Instruct	Microsoft Azure	Open	128k

LFM 40B	Liquid AI	Proprietary	32k

Solar Mini	Upstage	Open	4k

DBRX Instruct	Databricks	Open	33k

MiniMax-Text-01	MiniMax	Open	4m

Llama 3.1 Nemotron Instruct 70B	NVIDIA	Open	128k
Llama 3.3 Nemotron Super 49B v1	NVIDIA	Open	128k

Llama 3.1 Tulu3 405B	Allen Institute for AI	Open	128k

Reka Flash (Sep '24)	Reka AI	Proprietary	128k
Reka Core	Reka AI	Proprietary	128k
Reka Flash (Feb '24)	Reka AI	Proprietary	128k
Reka Edge	Reka AI	Proprietary	128k
Reka Flash 3	Reka AI	Open	128k

Command-R+ (Aug '24)	Cohere	Open	128k
Command-R+ (Apr '24)	Cohere	Open	128k
Command-R (Mar '24)	Cohere	Open	128k
Command-R (Aug '24)	Cohere	Open	128k
Command A	Cohere	Open	256k
Aya Expanse 32B	Cohere	Open	128k
Aya Expanse 8B	Cohere	Open	8k

Jamba 1.5 Large	AI21 Labs	Open	256k
Jamba 1.5 Mini	AI21 Labs	Open	256k
Jamba 1.6 Mini	AI21 Labs	Open	256k
Jamba 1.6 Large	AI21 Labs	Open	256k
Jamba Instruct	AI21 Labs	Proprietary	256k

Arctic Instruct	Snowflake	Open	4k

Qwen2.5 Max	Alibaba	Proprietary	32k
Qwen2.5 Instruct 72B	Alibaba	Open	131k
Qwen2.5 Coder Instruct 32B	Alibaba	Open	131k
Qwen Turbo	Alibaba	Proprietary	1m
Qwen2 Instruct 72B	Alibaba	Open	131k
QwQ 32B	Alibaba	Open	131k
Qwen1.5 Chat 110B	Alibaba	Open	32k
Qwen2.5 Instruct 32B	Alibaba	Open	128k
Qwen2.5 Coder Instruct 7B	Alibaba	Open	131k
Qwen Chat 72B	Alibaba	Open	34k
QwQ 32B-Preview	Alibaba	Open	33k

Yi-Large	01.AI	Proprietary	32k

OpenChat 3.5 (1210)	OpenChat	Open	8k