llm-benchmarks
githubdrose.io

fireworks Provider Benchmarks


Measured over the last 30 days

Models Tracked
20
Avg Tokens / Second
45.81
Avg Time to First Token (ms)
0.00
Last Updated
Aug 8, 2026

All fireworks models

20 tracked · fastest first
fireworks site
ModelAvg Toks/SecMinMaxAvg TTF (ms)
GPT-oss-120b98.405.87211.000.00
GPT-oss-20b74.4014.60158.000.00
MiniMax-M2.764.608.23136.000.00
qwen3-reranker-8b63.5011.1086.300.00
MiniMax-M359.109.40113.000.00
nemotron-3-ultra-550b-a55b59.109.2184.700.00
kimi-k2p7-code-fast55.9034.9085.400.00
qwen3p7-plus50.5012.0077.100.00
kimi-k2p6-turbo50.1013.3094.700.00
glm-5p2-fast49.001.23101.000.00
Kimi-K2.638.301.7779.800.00
DeepSeek-V4-Flash34.005.0361.100.00
Kimi-K2.7-Code33.101.5665.300.00
GLM-5.232.303.52109.000.00
DeepSeek-V4-Pro31.102.1549.700.00
kimi-k3-fast30.901.1749.200.00
Kimi-K330.501.7457.000.00
DeepSeek-V4-Flash-073129.001.1073.100.00
glm-5p128.102.4763.100.00
Inkling4.304.304.300.00

Frequently Asked Questions

Which fireworks model is fastest?
Based on recent tests, GPT-oss-120b shows the highest average throughput among tracked fireworks models.
How many recent measurements feed this dashboard?
This provider summary aggregates 4327 individual prompts measured across 3811 monitoring runs over the past month.