OpenAI compatible API. Attested gateway. Public status.

DeepSeek: DeepSeek V4 Pro Performance

TrustedRouter performance signals and provider route posture for DeepSeek: DeepSeek V4 Pro.

Verify gateway
1 URLbase_url migration
100smodels and routes
0prompt logs by default

deepseek/deepseek-v4-pro

open weights Performance

All models

AI IQ IQ 109 #28 public AI IQ rank for deepseek-v4-pro
View AI IQ profile

Measured performance

Continuously sampled p50/p95 time-to-first-token (TTFT), time-to-first-byte (TTFB), throughput, and success rate for DeepSeek: DeepSeek V4 Pro — unsupported route and probe-configuration rows are separated from provider downtime, and no prompt or output content is stored.

Providerp50 TTFTp95 TTFTp50 TTFBThroughputUptimeConfig excludedSamples
fireworks 2044 ms 15099 ms 2044 ms 90.91% 11
novita 2233 ms 6861 ms 2232 ms 100.00% 2
crusoe 5620 ms 18270 ms 5619 ms 88.89% 18
deepseek 5713 ms 16504 ms 5712 ms 100.00% 138
tinfoil 6351 ms 13705 ms 6351 ms 100.00% 44
siliconflow 7176 ms 16934 ms 7176 ms 100.00% 33
gmi 7279 ms 17710 ms 7279 ms 98.65% 74
baseten 9542 ms 24892 ms 9541 ms 82 tok/s 100.00% 43
wafer 0.00% 26
parasail 0.00% 4 probe_config_error 2

Full provider & model leaderboard.

Provider diversity

20 routes.

More routes give the auto router more room to fail over around provider 429 and 5xx responses.

Streaming

Gateway overhead is measured separately.

Public status separates TLS/health overhead from full model latency so slow LLMs do not inflate the router metric.

Status

Metadata rollups.

Status samples store latency, outcome, provider, model, route, cost, and region metadata only.

Sign in

Choose a sign in method.