Datasets:

pixeloffice
/

llm-smartrouter-benchmark

query_type stringclasses 6 values	target_model stringclasses 5 values	ttft_ms int64 160 410	total_latency_ms int64 701 1.98k	token_compression_pct stringclasses 7 values	cost_per_1m_tokens float64 0.08 3	status stringclasses 1 value
code_generation	deepseek/deepseek-chat	185	940	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	192	890	38%	0.14	200_ok
deep_reasoning	sference/thinkingcap-qwen3.6-27b	410	1,650	42%	0.28	200_ok
fast_chat	deepseek/deepseek-chat	175	720	29%	0.14	200_ok
fact_retrieval	pixeloffice/blun-auto	160	820	55%	0.12	200_ok
complex_refactor	bedrock/claude-sonnet-4-5	320	1,980	30%	3	200_ok
multi_file_analysis	google/gemini-2.5-flash	220	1,120	40%	0.075	200_ok
fast_chat	deepseek/deepseek-chat	161	701	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	162	702	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	163	703	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	164	704	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	165	705	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	166	706	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	167	707	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	168	708	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	169	709	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	170	710	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	171	711	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	172	712	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	173	713	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	174	714	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	175	715	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	176	716	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	177	717	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	178	718	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	179	719	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	180	720	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	181	721	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	182	722	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	183	723	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	184	724	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	185	725	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	186	726	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	187	727	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	188	728	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	189	729	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	190	730	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	191	731	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	192	732	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	193	733	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	194	734	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	195	735	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	196	736	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	197	737	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	198	738	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	199	739	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	160	740	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	161	741	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	162	742	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	163	743	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	164	744	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	165	745	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	166	746	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	167	747	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	168	748	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	169	749	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	170	750	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	171	751	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	172	752	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	173	753	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	174	754	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	175	755	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	176	756	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	177	757	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	178	758	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	179	759	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	180	760	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	181	761	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	182	762	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	183	763	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	184	764	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	185	765	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	186	766	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	187	767	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	188	768	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	189	769	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	190	770	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	191	771	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	192	772	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	193	773	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	194	774	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	195	775	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	196	776	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	197	777	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	198	778	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	199	779	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	160	780	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	161	781	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	162	782	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	163	783	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	164	784	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	165	785	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	166	786	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	167	787	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	168	788	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	169	789	35%	0.14	200_ok
deep_reasoning	deepseek/deepseek-chat	170	790	35%	0.14	200_ok
fact_retrieval	deepseek/deepseek-chat	171	791	35%	0.14	200_ok
code_generation	deepseek/deepseek-chat	172	792	35%	0.14	200_ok
fast_chat	deepseek/deepseek-chat	173	793	35%	0.14	200_ok

End of preview. Expand in Data Studio

LLM SmartRouter Latency & Cost Benchmark (2026)

Empirical performance benchmark dataset comparing direct model endpoints (OpenAI, Anthropic, Google) against the PixelRouter / BLUN SmartRouter proxy layer (https://api.pixeloffice.eu/v1).

Schema

query_type: Task classification (code_generation, deep_reasoning, fast_chat, fact_retrieval).
target_model: Routed model identifier.
ttft_ms: Time-To-First-Token in milliseconds.
total_latency_ms: End-to-end response completion time.
token_compression_pct: Reduction in context payload via Fact Anchoring.
cost_per_1m_tokens: Cost in USD compared to raw proprietary APIs.
status: Execution health check (200_ok).

Reference

Maintained by Pixel Office EU and available via pip install pixeloffice-router or npm install @pixeloffice-eu/router.

Downloads last month: 59