40% cheaper AI tokens
Frontier open models, optimized for speed and quality.
$5 credits to start · Zero Data Retention
Kimi-K3.1-Pro1M context · text, imageInput
Muse Spark Open1M context · text, image, video, audio, PDFInput
GLM-5.31M context · textInput
DeepSeek-V4.1-Flash-10171M context · text, imageInputGLM-5.3-Flash1M context · text, image, videoInput
Kimi-K3
Bring your own keysSwitch by changing one URL
Point your existing client at SkyPilot Tokens. Nothing else to be changed.
- OpenAI and Anthropic compatible
- Works with both SDKs, and with Claude Code and OpenCode.
- Streaming, tool calls, JSON mode
- No changes to your existing agentic workflows.
- One key for every workload
- Agents, evals, and batch jobs on the same key.
from openai import OpenAI client = OpenAI( base_url="https://tokens.skypilot.ai/v1", api_key="YOUR_SKYPILOT_TOKENS_KEY", ) response = client.responses.create( model="meta-models/Muse-Spark-Open-816B-A42B", input="Triage these support tickets.", )
Calculate how much open models save you
| List price | GPT-5.6 Sol | GLM-5.3 |
|---|---|---|
| Cached input / M | $0.40 | $0.156 |
| Input / M | $4.00 | $0.84 |
| Output / M | $20.00 | $2.64 |
Moving to GLM-5.3 (suggested)| How the savings compound | Year 1 | Year 2 | Year 3 |
|---|---|---|---|
| Today's vendor | $1.8M | $2.7M | $4.0M |
| SkyPilot Tokens | $284k | $426k | $638k |
| Saved | $1.5M | $2.3M | $3.4M |
Show the math
- Blended GPT-5.6 Sol price80% input, 70% of it cached$5.18 / M
- Tokens a month$150k ÷ $5.1828,935M
- Blended GLM-5.3 pricesame token mix$0.817 / M
- SkyPilot Tokens a month28,935M × $0.817$24k
- Saved a month$150k − $24k$126k
- Vendor prices
- Public list prices per 1M tokens, read 2026-09-13; GPT-5.6 Sol: promotional price through Nov 21, 2026.
- Our prices
- The prices on this page.
- Token mix
- Agentic workflows: 80% of tokens are input and 70% of input is cached. Your spend sets the volume, and both sides serve the same traffic.
- Suggested model
- Our capability match, not a guarantee.
Running production workloads? Book a call
We take privacy seriously
Requests are encrypted on every hop, and nothing you send is stored.
- Zero Data Retention
- Prompts and outputs are never stored, by SkyPilot or on the fleet.
- Caches salted per account
- One customer’s cached prompts are never reused for another.
- Both sides stay anonymous
- Buyers can’t identify fleets, and GPU owners can’t identify buyers.
- SOC 2 Type II
- Report available on our Trust Center.
- Your appOpenAI or Anthropic SDK
- TLSNo prompts or outputs stored or logged
- SkyPilot TokensGateway · every request and response anonymized
- mTLS tunnel · certs rotated daily
- SkyPilot-verified fleetModel server on a SkyPilot hardened image
Why it costs 40% less
Served on idle reserved GPUs managed by SkyPilot Platform.
- Run on GPUs already paid for
- Owners already paid for them, so idle hours cost nothing extra.
- Every GPU fleet verified
- Each fleet is a contracted and verified GPU owner.
- Optimized for speed and quality
- Optimized for B200 and B300; matched model provider’s own API quality.
“On launch day, we scaled to 10s of billions of tokens per hour without degradation. SkyPilot is a critical partner in serving all of our models.”
FAQ
How can SkyPilot Tokens be 40% cheaper?
SkyPilot Tokens is built on a novel approach: serving tokens on SkyPilot-managed GPUs from verified customers. These customers already use SkyPilot Platform to schedule their GPU fleets, and they can opt in to allow SkyPilot Tokens serving on idle nodes.
Serving on idle GPUs (with dedicated fallback capacity) means we can offer 40% cheaper cost than most inference providers.
What happens when a GPU owner takes GPUs back?
In-flight requests automatically get re-routed to other available GPU nodes, including other idle GPU nodes or dedicated fallback capacity. Latency can increase briefly, but the endpoint keeps serving.
Is SkyPilot Tokens secure?
Yes. We take security and privacy design very seriously. Zero Data Retention (ZDR) is always on. Your requests and responses are not stored by us or any third party. Refer to the privacy section above for technical details. Reach out to our team for in-depth discussions.
Where do my requests run?
Requests run on GPU fleets from SkyPilot Platform’s verified customers, and can fall back to our dedicated GPUs or top-tier partner inference providers.
| Backend | What it is | Data handling | Traffic |
|---|---|---|---|
| Verified SkyPilot Platform customer fleets | Idle GPUs in verified customers' fleets, serving a model server we deploy and operate | Zero Data Retention; per-account cache salt; requests arrive over an mTLS tunnel | Most |
| SkyPilot dedicated nodes | Our own single-tenant nodes on AWS, OCI, CoreWeave, Together AI, and Nebius, running the same model server | Zero Data Retention; per-account cache salt | Most |
| Nebius Token Factory | Partner inference API | Zero Data Retention enabled for our account | Spill-over only |
| Together AI | Partner inference API | Zero Data Retention enabled for our account | First for some models |
| Baseten | Partner inference API | Zero Data Retention enabled for our account | Spill-over only |
Every enterprise customer that joins the fleet goes through technical verification and a legal contract. Model servers are deployed by SkyPilot Tokens and use a limited set of specific hardened images.
Which models are available?
Kimi-K3.1-Pro, Muse Spark Open, GLM-5.3, GLM-5.3-Flash, Kimi-K3, and DeepSeek-V4.1-Flash-1017 today, with more open models continuously added. You can also bring your own keys (BYOK) to access OpenAI GPT or Anthropic Claude models through the same endpoint.