Wafer — pricing & performance
Models served
12
of 1,191 tracked models
Mean output speed
95
tokens per second, averaged over
12 measured endpoints
12 measured endpoints
Mean latency
1.1s
time to first token, averaged over
12 measured endpoints
12 measured endpoints
Cheapest blended
$0.069
per 1M tokens at a 3:1
input:output blend
input:output blend
Output speed on Wafer
Output tokens per second for all 12 models on Wafer with published throughput telemetry, the most recent figure OpenRouter publishes for each host. Higher is better. We do not run these measurements ourselves.
The Known Good
12 models on Wafer
Full leaderboard →
| Model | Creator | Input $/M | Output $/M | Blended | Speed | TTFT | Context |
|---|---|---|---|---|---|---|---|
DeepSeek: DeepSeek V4 Flash 0423
|
DeepSeek | $0.035 | $0.17 | $0.069 | 38 | 0.8s | 1.05M |
|
DeepSeek: DeepSeek V4 Flash 0731
|
DeepSeek | $0.13 | $0.23 | $0.15 | 112 | 0.5s | 1.05M |
|
DeepSeek: DeepSeek V4 Pro 0423
|
DeepSeek | $1.20 | $2.40 | $1.50 | 28 | 5.1s | 1.05M |
|
DeepSeek: DeepSeek V4 Pro 0813
|
DeepSeek | $0.15 | $3.40 | $0.96 | 114 | 0.5s | 1.05M |
|
DeepSeek: DeepSeek V4.1 Flash
|
DeepSeek | $0.045 | $0.40 | $0.13 | 82 | 0.5s | 1.05M |
MoonshotAI: Kimi K3
|
Moonshot AI | $0.94 | $15.00 | $4.46 | 54 | 0.6s | 1.05M |
NVIDIA: Nemotron 3.5 Lightning
|
NVIDIA | $0.050 | $0.20 | $0.087 | 356 | 0.9s | 262k |
Qwen: Qwen3.8 27B
|
Alibaba | $0.024 | $2.90 | $0.74 | 42 | 0.8s | 262k |
Z.ai: GLM 5.1
|
Z.ai | $1.00 | $3.20 | $1.55 | 60 | 1.2s | 203k |
|
Z.ai: GLM 5.2
|
Z.ai | $0.084 | $3.60 | $0.96 | 121 | 0.5s | 1.05M |
|
Z.ai: GLM 5.3
|
Z.ai | $0.049 | $3.39 | $0.88 | 85 | 0.6s | 1.05M |
|
Z.ai: GLM 5.3 Flash
|
Z.ai | $0.070 | $0.30 | $0.13 | 43 | 0.7s | 1.05M |
The Known Good
Source. Prices, endpoint availability and throughput for Wafer 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.