Snapshot
Continue to About ↓- What it does
- Wafer runs AI agents that profile a model-serving workload, search across batching, decoding, quantization, engines, kernels and hardware, then deploys the fastest configuration that still meets correctness and reliability targets and keeps re-optimizing as traffic, models and hardware change.
More detail
- It offers this as dedicated, quote-based deployments and as a serverless, OpenAI- and Anthropic-compatible API for open models billed per token.
- Best for
- AI product, voice and infrastructure teams serving open models who need lower latency, stable performance under rising requests per minute, or better throughput on specific accelerators such as AMD GPUs.
- Pricing
- Three economic surfaces.
More detail
- Serverless is billed per token with published per-model input, output (including reasoning tokens) and cache-read rates
- for example, GLM-5.2 is listed at $1.20 per million input tokens, $4.10 per million output tokens and $0.20 per million cache-read tokens, funded through
- Main limitation
- Buyers who need a published fixed rate card for dedicated deployments, a subscription tier without a three-request concurrency cap, or refunds on prepaid subscription fees.
- Record
- Static profile
- Reviewed
- Evidence
- Public web research
- Machine access
- Structured profile ↗
About
- Wafer runs AI agents that profile a model-serving workload, search across batching, decoding, quantization, engines, kernels and hardware, then deploys the fastest configuration that still meets correctness and reliability targets and keeps re-optimizing as traffic, models and hardware change.
- It offers this as dedicated, quote-based deployments and as a serverless, OpenAI- and Anthropic-compatible API for open models billed per token.
What it helps with
- High and unpredictable inference latency that degrades as request volume rises.
- Inference cost inflated by extra application layers such as prefetching or caching that become unnecessary with lower la
- Slow time-to-first-token in voice and conversational agents.
- Poor performance of open models on AMD accelerators.
- Manual, point-in-time serving tuning that goes stale as traffic, models and hardware change.
Capabilities
A practical view of what is available and where limitations apply.
| Offering | Available | Detail |
|---|---|---|
| Continual inference optimization | Yes | Wafer says it learns how a workload behaves and keeps re-optimizing the serving stack after deployment. |
| Across-the-stack search | Yes | Generates and measures candidate configurations across batching, decoding, quantization, engines, kernels and hardware. |
| Correctness and reliability gate | Yes | Every candidate must preserve correctness and meet reliability targets before it can ship. |
| Dedicated deployments | Yes | Endpoints built around a customer's model, traffic shape and SLO; pricing is quote-based. |
| Serverless inference API | Yes | Per-token billing for open models via app.wafer.ai. |
| OpenAI and Anthropic API compatibility | Yes | Supports /v1/chat/completions, /v1/responses and /v1/messages. |
| Zero Data Retention by default | Yes | Applied on every serverless request without a header; model support varies. |
| AMD GPU support | Yes | Works across NVIDIA and AMD GPUs, with a published Kimi 2.5 case on AMD MI355X. |
| Third-party independent benchmarks | Unknown | No independent reviews or benchmarks were found; performance figures in the sources are first-party or first-party-published customer quotes. |
| Published dedicated-deployment rate card | No | Dedicated pricing is bespoke and undisclosed in the sources. |
Bot trust
How confidently AI systems can verify this profile from public evidence.
—of 100
ChatGPT—/100
Claude—/100
This profile has not completed a Bot Score assessment yet.
Evidence
Where this record came from, when it was reviewed, and how confidently it can be used.
- Verification
- Scribble researched these facts from the linked public sources.
- The owner has not manually confirmed every field.
- Last reviewed
- Primary source
- wafer.ai
Technical profile details
Record ID
64a27e77-39ec-4a71-ae10-4a349e074fe7Canonical URLhttps://profiles.scribble.network/listing/wafer-64a27e77