The Known Good Updated 8 Oct 2026 Subscribe
Providers· Inference host· 12 tracked models

Relace — pricing & performance

Compare models All providers Data sources
Models served
12
of 1,191 tracked models
Mean output speed
82
tokens per second, averaged over
11 measured endpoints
Mean latency
1.0s
time to first token, averaged over
11 measured endpoints
Cheapest blended
$0.15
per 1M tokens at a 3:1
input:output blend
Output speed on Relace

Output tokens per second for the 11 of 12 models on Relace with published throughput telemetry, the most recent figure OpenRouter publishes for each host. The full table below lists all 12, measured or not. Higher is better. We do not run these measurements ourselves.

⇩
113
104
99
98
92
86
85
74
56
54
45
The Known Good
12 models on Relace
Full leaderboard →
Model Creator Input $/M Output $/M Blended Speed TTFT Context
DeepSeek: DeepSeek V4 Flash 0423 DeepSeek $0.006 $1.28 $0.32 98 0.5s 1.05M
DeepSeek: DeepSeek V4 Flash 0731 DeepSeek $0.018 $1.28 $0.33 45 0.7s 1.05M
DeepSeek: DeepSeek V4 Pro 0423 DeepSeek $0.28 $4.20 $1.26 86 1.0s 1.05M
DeepSeek: DeepSeek V4 Pro 0813 DeepSeek $0.13 $8.00 $2.10 113 1.0s 1.05M
DeepSeek: DeepSeek V4.1 Flash DeepSeek $0.016 $1.20 $0.31 104 1.0s 1.05M
MoonshotAI: Kimi K3 Moonshot AI $0.81 $13.00 $3.86 56 1.5s 1.05M
Relace: Relace Apply 3 relace $0.85 $1.25 $0.95 92 1.6s 256k
Relace: Relace Search relace $1.00 $3.00 $1.50 — — 256k
Xiaomi: MiMo-V2.6-Flash Xiaomi $0.040 $1.28 $0.35 54 0.4s 1.05M
Z.ai: GLM 5.2 Z.ai $0.070 $4.30 $1.13 85 1.2s 1.05M
Z.ai: GLM 5.3 Z.ai $0.039 $4.40 $1.13 99 1.1s 1.05M
Z.ai: GLM 5.3 Flash Z.ai $0.040 $0.50 $0.15 74 0.8s 1.05M
The Known Good
Source. Prices, endpoint availability and throughput for Relace are ingested from OpenRouter; model metadata is enriched from models.dev. A blank cell means the figure was not published for that endpoint — it never means zero. Blended price is (3 × input + output) ÷ 4, the same 3:1 blend used everywhere on this site; see methodology. Quality scores are not provider-specific: a model scores the same wherever it is hosted, so quality lives on the model pages.