Updated Sep 13, 10:54 PM UTC
Model Data
See which models are winning real usage, how the mix is shifting, and what that means for cost.
Top Models. Usage of models across OpenCode.
010203
muse-spark-1.3-contributor35T
Meta+109%
deepseek-v4-flash21T
DeepSeek-15%
deepseek-v4.1-flash19T
DeepSeekNew
040506070809101112131415161718
mimo-v2.59.2T
Xiaomi-15%
muse-spark-1.2-contributor4.3T
Meta-60%
glm-5.3-flash3.4T
Zhipu-7%
nemotron-3-ultra2.4T
NVIDIA-14%
omen-alpha1.2T
+70%
deepseek-v4-flash-vision-exp1.0T
DeepSeek-43%
deepseek-v4-pro722B
DeepSeek-31%
gpt-5.6-luna559B
OpenAI+3%
nemotron-3.5-lightning481B
NVIDIA-44%
qwen3.8-flash455B
-25%
ling-3.0-flash-fin402B
-3%
minimax-m3296B
MiniMax-11%
mimo-v2.5-pro173B
Xiaomi-10%
kimi-k2.7-code164B
Moonshot-3%
glm-5.2135B
Zhipu-30%
010203040506070809101112131415161718
muse-spark-1.3-contributor35T
Meta+109%
deepseek-v4-flash21T
DeepSeek-15%
deepseek-v4.1-flash19T
DeepSeekNew
mimo-v2.59.2T
Xiaomi-15%
muse-spark-1.2-contributor4.3T
Meta-60%
glm-5.3-flash3.4T
Zhipu-7%
nemotron-3-ultra2.4T
NVIDIA-14%
omen-alpha1.2T
+70%
deepseek-v4-flash-vision-exp1.0T
DeepSeek-43%
deepseek-v4-pro722B
DeepSeek-31%
gpt-5.6-luna559B
OpenAI+3%
nemotron-3.5-lightning481B
NVIDIA-44%
qwen3.8-flash455B
-25%
ling-3.0-flash-fin402B
-3%
minimax-m3296B
MiniMax-11%
mimo-v2.5-pro173B
Xiaomi-10%
kimi-k2.7-code164B
Moonshot-3%
glm-5.2135B
Zhipu-30%
Unique Users. Daily unique OpenCode users by model.
Weekly Retention. Weekly users returning the next week.
- 01muse-spark-1.2-contributor85.4%5.5K
- 02qwen3.8-flash82.1%363
- 03hy4-preview81.1%159
- 04deepseek-v4-flash-vision-exp74.8%7.2K
- 05hy373.0%2.6K
- 06mimo-v2.572.9%22K
- 07qwen3.5-plus72.5%280
- 08kimi-k2.570.7%834
- 09deepseek-v4-flash69.1%455K
- 10minimax-m2.768.2%3.3K
- 11glm-5.3-flash67.5%24K
- 12minimax-m2.565.9%337
- 13glm-565.5%800
- 14mimo-v2.5-pro64.9%7.7K
- 15kimi-k2.664.8%8.9K
Session Cost. Average cost per session.
COST / SESSION
TOKENS / SESSION
Cost/Session$0.0015
Tokens/Session667K
Token Cost. Price per 1M tokens.
Input$0.30
Output$1.20
Cached$0.0060
Cache Ratio. Share of input tokens served from cache.
Cache Ratio95%
Cached33T
Uncached1.9T
Model Comparisons. Popular model pairs from the leaderboard.
Top two by recent usagemuse-spark-1.3-contributor vs deepseek-v4-flashmuse-spark-1.3-contributor·deepseek-v4-flashLeader vs challengermuse-spark-1.3-contributor vs deepseek-v4.1-flashmuse-spark-1.3-contributor·deepseek-v4.1-flashAdjacent leaderboard pairdeepseek-v4-flash vs deepseek-v4.1-flashdeepseek-v4-flash·deepseek-v4.1-flashTop model alternativedeepseek-v4.1-flash vs mimo-v2.5deepseek-v4.1-flash·mimo-v2.5
