LLMs By Max Output Token - Feb 2025

This analysis shows the evolution of maximum output tokens across different AI language models and companies. The visualization uses a logarithmic scale to better display the wide range of token values (4,000 to 100,000) and includes trend lines to show each company's progression over time. Models with 32K+ tokens are labeled directly on the plot for easy reference.

Visualization

Max Output Tokens / Time

Chart 1 With Connections To Show Per-Company Evolution

This chart really highlights how OpenAI, with their recently released o1 series, have brought the first significant advances to the max output tokens parameter.

While models like the Gemini series have amazing context sizes of 1M+ tokens, their max single output remains a very diminuitive 8,192 tokens (0.008%!).

The maximum output size is highly relevant when debugging or creating long code projects in which even with agentic coding capabilities that only update part of a long file at a time, the constraint can make write-to-file type commands non-functional due to continuous output constraints:

Repository Structure

• data/: Contains the primary dataset and source data
  • data-sources/: Original data collected from each company's documentation
    • anthropic/: Anthropic model data
    • cohere/: Cohere model data
    • google/: Google model data
    • openai/: OpenAI model data
• plots/: Contains visualization outputs
• 080225/: Contains the latest analysis and findings

Data Table (Sorted by Maximum Output Tokens)

Model Name	Max Output Tokens	Company	Launch Date	Max Input Tokens	Max Output as % of Max Input	Release Type
o1	100000	OpenAI	2024-11-06	1000000	10.00	Fixed
o1-mini	65536	OpenAI	2024-11-06	1000000	6.55	Fixed
Deep Seek Reasoner	32768	Deep Seek	2025-01-20	32768	100.00	Rolling
o1-preview	32768	OpenAI	2024-11-06	1000000	3.28	Fixed
chatgpt-4o-latest	16384	OpenAI	2024-11-06	128000	12.80	Fixed
gpt-4o-mini	16384	OpenAI	2024-11-06	128000	12.80	Fixed
gpt-4o	16384	OpenAI	2024-11-06	128000	12.80	Fixed
Gemini 2.0 Flash	8192	Google	2025-02-05	1000000	0.82	Fixed
Gemini 2.0 Flash-Lite Preview	8192	Google	2025-02-05	1000000	0.82	Fixed
gpt-4-0314	8192	OpenAI	2023-03-14	8192	100.00	Fixed
gpt-4-0613	8192	OpenAI	2023-06-13	8192	100.00	Fixed
gpt-4	8192	OpenAI	2023-03-14	8192	100.00	Fixed
Claude 3.5 Sonnet	8192	Anthropic	2024-06-20	200000	4.10	Fixed
Claude 3.5 Haiku	8192	Anthropic	2024-10-22	200000	4.10	Fixed
Gemini 1.5 Pro	8192	Google	2024-05-23	1000000	0.82	Fixed
Gemini 1.5 Flash 8B	8192	Google	2024-10-08	1000000	0.82	Fixed
Gemini 1.5 Flash	8192	Google	2024-05-14	1000000	0.82	Fixed
Deep Seek V3	8000	Deep Seek	2025-01-20	32768	24.41	Rolling
Claude 3 Haiku	4096	Anthropic	2024-03-07	200000	2.05	Fixed
Claude 3 Opus	4096	Anthropic	2024-03-14	200000	2.05	Fixed
gpt-4-turbo	4096	OpenAI	2023-11-06	128000	3.20	Fixed
gpt-4-turbo-preview	4096	OpenAI	2023-11-06	128000	3.20	Fixed
Claude 3 Sonnet	4096	Anthropic	2024-03-14	200000	2.05	Fixed
command-r-plus-08-2024	4000	Cohere	2024-08-01	128000	3.13	Fixed
command-r7b-12-2024	4000	Cohere	2024-12-01	128000	3.13	Fixed

Author

Daniel Rosehill
(public at danielrosehill dot com)

Licensing

This repository is licensed under CC-BY-4.0 (Attribution 4.0 International) License

Summary of the License

The Creative Commons Attribution 4.0 International (CC BY 4.0) license allows others to:

• Share: Copy and redistribute the material in any medium or format.
• Adapt: Remix, transform, and build upon the material for any purpose, even commercially.

The licensor cannot revoke these freedoms as long as you follow the license terms.

License Terms

• Attribution: You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.
• No additional restrictions: You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.

For the full legal code, please visit the Creative Commons website.