Last refreshed 2026-06-04. Next refresh: weekly.
Why use Kimi K2 Instruct on NVIDIA NIM?
NVIDIA NIM offers Kimi K2 Instruct with competitive pricing. NVIDIA NIM is NVIDIA's deployment platform for GPU-accelerated inference microservices.
Compare Kimi K2 Instruct across 5 providers to find the best fit for your use caseInput / 1M
-
Output / 1M
-
Cache
Not sourced
Batch
Not sourced
Setup recipe
Docs fallbackInstall
Use the provider REST API or SDKAuth
Create a provider API keyCall
model: moonshotai/kimi-k2-instructModel ID
moonshotai/kimi-k2-instructRequest example
Curated snippets for this provider are not sourced yet. Use NVIDIA NIM documentation with model ID
moonshotai/kimi-k2-instruct.Gotchas
- Use provider model ID "moonshotai/kimi-k2-instruct", not the LLMReference slug "kimi-k2-instruct".
Compare Kimi K2 Instruct Across Providers
| Provider | Input (per 1M) | Output (per 1M) |
|---|---|---|
| Fireworks AI | $0.60 | $2.50 |
| Together AI | $1.20 | $4.50 |
| NVIDIA NIM | — | — |
| Vercel AI Gateway | $0.57 | $2.30 |
| Novita AI | $0.57 | $2.30 |
Pricing
| Type | Rate |
|---|---|
| GPU Hour Rate | $1.00/GPU·hr |
Capabilities
ReasoningStructured Outputs
About Kimi K2 Instruct
Kimi K2 Instruct is an instruction-tuned language model from Moonshot AI, available via Fireworks AI.
FAQ
What is the context window for Kimi K2 Instruct on NVIDIA NIM?
Kimi K2 Instruct supports a 131k token context window on NVIDIA NIM.
How does NVIDIA NIM compare to other Kimi K2 Instruct providers?
Kimi K2 Instruct is available from 5 providers. The cheapest input pricing is $0.57/1M tokens from Vercel AI Gateway.
What API model ID do I use for Kimi K2 Instruct on NVIDIA NIM?
Use the model ID moonshotai/kimi-k2-instruct when calling NVIDIA NIM's API.
Who created Kimi K2 Instruct?
Kimi K2 Instruct was created by Moonshot AI as part of the Kimi K2 model family.
Is Kimi K2 Instruct open source?
Kimi K2 Instruct is open source under MIT according to the seed data.
Model Specs
Released2025-09-05
Parameters1T total, 32B active (MoE)
Context131k
ArchitectureDecoder Only