> Lambda AI
Lambda sells NVIDIA GPU cloud rather than a token-metered API. Self-serve instance pricing is per GPU-hour: NVIDIA B200 SXM6 at $6.69, H100 SXM at $3.99, A100 SXM 80GB at $2.79, A100 SXM 40GB at $1.99, GH200 96GB at $2.29, A10 at $1.29, A6000 at $1.09 and Tesla V100 at $0.79, all plus applicable sales tax/VAT/GST. 1-Click Clusters scale from 16 to 2,000+ GPUs on 2-week-to-1-year terms at $8.87-$9.86 per GPU-hour for HGX B200 and $5.54-$6.16 for H100. Lambda's own Inference API is being retired: the official page states that "as the Inference API winds down" customers should continue by deploying and scaling models on NVIDIA GPU instances.
// pros
- Instance catalogue spans cost tiers from $0.69/GPU/hour (Quadro RTX 6000) to $6.69 (B200 SXM6), with VRAM, vCPU, RAM and local SSD published per SKU for each node size.
- Cluster pricing falls with both scale and term: HGX B200 from $9.86/GPU-hour at 16 GPUs to $8.87 at 256+, and H100 from $6.16 at 16 GPUs to $5.54 at 256, on 2-week to 1-year reservations.
- Standard instances are self-serve and first-come with no long-term commitment, so no sales cycle is needed for single-node B200, H100, A100 or GH200 work.
- Public Cloud supports 16x to 512x H100/B200 clusters with a choice of Managed Kubernetes, Preinstalled Kubernetes or Slurm, per the official docs.
- Multi-GPU instances are InfiniBand-connected and come with full SSH access, which suits fine-tuning and distributed training in a way that token APIs do not.
// cons
- No free tier or trial credit is published on the pricing page, and every rate is exclusive of sales tax/VAT/GST.
- The first-party Inference API is winding down, so there is no serverless per-token text endpoint to fall back on; inference must be self-hosted on rented GPUs.
- Hourly GPU billing means idle capacity is chargeable, which is structurally more expensive than per-token serverless for bursty or low-volume inference.
- Private Cloud, aimed at single-tenant clusters of 1,000+ GPUs, is a contact-sales tier with no published rates.
Lambda is a reasonable fit for training, fine-tuning and sustained self-hosted inference where a fixed hourly rate and SSH access matter more than elasticity; for spiky low-volume traffic, pair it with a per-token serverless provider, since Lambda no longer offers a first-party inference endpoint.
written by Mike Fleming, Uprouter editorial · verified against official sources September 16, 2026
No free tier
This provider does not offer a free tier as of our last verification (Sep 16, 2026). The cheapest way to try it is pay-as-you-go — input from Free per 1M tokens. If the provider adds a free tier, submit a correction and we’ll re-verify.
| Plan | Type | Monthly | Input /1M | Output /1M | Note |
|---|---|---|---|---|---|
| Pay-as-you-go | payg | — | — | — | Billed per 1M tokens; rates depend on model size. |
Pricing normalized from public sources — always verify with the provider.
// model prices
| Model | Input /1M | Output /1M | Context |
|---|---|---|---|
| Meta: Llama 3.1 405B Instruct | Free | Free | 131k |
| Mistral Large | $0.0000 | $0.0000 | 128k |
$ Does Lambda AI have a free tier?
No. Lambda AI does not offer a meaningful free tier — entry requires payment (pay-as-you-go or a plan). We mark this explicitly rather than showing a misleading zero.
$ How much does Lambda AI cost per 1M tokens?
The cheapest model we track at Lambda AI is Free per 1M input tokens and Free per 1M output tokens. This is normalized from Lambda AI's published pricing — verify with the provider before purchasing, since prices change frequently.
$ Is Lambda AI safe to use?
Uprouter rates Lambda AI as low risk. Risk is moderate. Rates are published and capacity is self-serve, but billing is hourly and taxed on top, so idle GPU time is real cost and there is no free-tier hedge. The clearest directional risk is product retirement: Lambda's own inference page states the Inference API is winding down, so any integration built against that endpoint must migrate to self-hosted deployments on instances or clusters. No SLA, region list or data-retention terms appear on the pricing and docs pages fetched. Risk ratings are editorial, evidence-based, and never influenced by affiliate relationships.
Raw community sentiment, shown unblended. Votes never feed into objective data fields like price, risk, or status — and never into sort order.
No published reviews yet — be the first.
// more_informationprice history · live status · risk & confidence · code snippets
price_history
No recorded changes yet.
live_status
current: unknownNo probes recorded yet.
Uprouter runs its own probes; best-effort, not a guarantee. Probe results depend on our network location and cadence — for production, always use the provider’s own status page.
risk_&_confidence
Risk is moderate. Rates are published and capacity is self-serve, but billing is hourly and taxed on top, so idle GPU time is real cost and there is no free-tier hedge. The clearest directional risk is product retirement: Lambda's own inference page states the Inference API is winding down, so any integration built against that endpoint must migrate to self-hosted deployments on instances or clusters. No SLA, region list or data-retention terms appear on the pricing and docs pages fetched.
Confidence reflects how much verifiable evidence (official docs, probe history, community reports) backs this entry. Risk is our editorial assessment of operator reliability and terms — not financial advice.
code_snippets
# Lambda AI's own API — get a key from their dashboard
curl https://lambda.ai/chat/completions \
-H "Authorization: Bearer THEIR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "meta-llama/llama-3.1-405b-instruct",
"messages": [{ "role": "user", "content": "Hello" }]
}'Direct to the provider — Uprouter receives nothing; verify their current pricing before purchasing.
Ranked by neutral similarity signals only — live status, free-tier shape and free-credit proximity. Commercial relationships never influence this list.