Providers/Cloudflare
Cloudflare logo

Cloudflare

Inference providerdevelopers.cloudflare.com ↗Status ↗
MODELS
19
ENDPOINTS
19
MIN INPUT
$0.017
MEDIAN BLENDED
$0.66
AVG UPTIME
99.4%
CHEAPEST FOR
4 models
Models on Cloudflare · prices & measurements for models served by Cloudflare’s APIUSD / MTOK · VERIFIED 23 MINUTES AGO
MODELINPUTOUTPUTCONTEXTQUANTTTFTTPSSMOKEVS BEST ELSEWHERE
IBM Granite logoGranite 4.0 MicroIBM Granite$0.017$0.112131K—✕✕—only provider
Meta logoLlama 3.2 1B InstructMeta$0.027$0.20160K—0.5s°93° tok/s—only provider
Meta logoLlama 3.2 3B InstructMeta$0.051$0.33580K—0.58s°271° tok/s—$0.05 elsewhere
Z.ai logoGLM 4.7 FlashZ.ai$0.06$0.40131K—✕✕—$0.06 elsewhere
Google logoGemma 4 26B A4B Google$0.10$0.30256K—✕✕5/5$0.042 elsewhere
Meta logoLlama 3.1 8B InstructMeta$0.152$0.28732Kfp82.4s°14° tok/s3/5$0.02 elsewhere
Meta logoLlama 3.3 70B InstructMeta$0.293$2.2524Kfp84.6s°39° tok/s4/5$0.10 elsewhere
Z.ai logoGLM 5.3 FlashZ.ai$0.30$1.001.05M—1.1s°27° tok/s—$0.032 elsewhere
Mistral logoMistral Small 3.1 24BMistral$0.351$0.555128K—1.6s°31° tok/s—only provider
DeepSeek logoDeepSeek V4 Flash 0731DeepSeek$0.44$1.321.05M—0.68s°86° tok/s—$0.01 elsewhere
DeepSeek logoDeepSeek V4 Flash 0423DeepSeek$0.44$1.32384K————$0.017 elsewhere
Qwen logoQwen3.8 27BQwen$0.45$3.20262K—1.8s°52° tok/s—$0.024 elsewhere
Qwen logoQwen2.5 Coder 32B InstructQwen$0.66$1.0033K—2.2s°35° tok/s—only provider
Moonshot AI logoKimi K2.7 CodeMoonshot AI$0.95$4.00262K—1.6s°106° tok/s—$0.671 elsewhere
Moonshot AI logoKimi K2.6Moonshot AI$0.95$4.00262K—✕✕—$0.465 elsewhere
DeepSeek logoDeepSeek V4 Pro 0423DeepSeek$1.15$2.551.05M—1.4s°47° tok/s—$0.30 elsewhere
Z.ai logoGLM 5.2Z.ai$1.18$4.40262K—2.1s°62° tok/s—$0.03 elsewhere
DeepSeek logoDeepSeek V4 Pro 0813DeepSeek$1.32$3.961.05M—1.5s°31° tok/s—$0.158 elsewhere
Z.ai logoGLM 5.3Z.ai$1.40$4.401.05M—6.3s°80° tok/s—$0.039 elsewhere
TTFT/TPS: ModelIndex-measured medians (P50, trailing 21d) for Cloudflare’s endpoints · ° = measured via OpenRouter routing · ✕ = probed, no measurable stream · SMOKE: of 5 capability probes passed · methodology on each model page