Kimi K3 · regional API access

Use Kimi K3 from Russia through one compatible API

AnyModel exposes Kimi K3 through an OpenAI-compatible endpoint and an AnyModel prepaid balance. It is an independent route backed by connected Kimi Code capacity, not a direct Moonshot account or Kimi subscription.

Kimi K3256K contextCodex-compatiblePrepaid balance

Create a free API key See live pricing

What the route includes

Compatible API access, not a direct Kimi account

The model, account and billing layers are separate. AnyModel gives your application a compatible endpoint and one balance while the upstream Kimi route remains managed by the service.

Option	Key	Billing	Model switching
Kimi K3 through AnyModel	AnyModel key	AnyModel prepaid balance	GPT, Claude, Gemini, GLM and Kimi
Direct Moonshot API	Moonshot key	Moonshot account billing	Moonshot catalog
Kimi Code subscription	Kimi account	Subscription tier	Kimi Code products

API quickstart

Send a Kimi K3 request in three steps

1
Create an AnyModel key
Register, copy the key and confirm that Kimi K3 is available in the live catalog.
2
Use the OpenAI-compatible endpoint
Point the client at https://anymodel.org/v1 and send the exact model id kmc/k3.
3
Measure the real workload
Test context size, latency, patch quality and balance usage before relying on the route for a long autonomous job.

curl · Kimi K3 chat completion

curl https://anymodel.org/v1/chat/completions \
  -H "Authorization: Bearer $ANYMODEL_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"kmc/k3","messages":[{"role":"user","content":"Review this refactor plan."}]}'

Regional access, payment and limits

Requests authenticate with an AnyModel key, so you do not need to create or fund a separate Moonshot account. This does not grant a direct Kimi account or change Moonshot's terms and regional policies.

The checkout shows payment methods currently available to your account. Cards, local transfers, SBP, crypto and settlement currencies depend on the active provider and should never be assumed from a marketing page.

Kimi K3 has a 256K context limit on this route. Long agent sessions can also consume capacity across many sequential requests, so watch both raw context and measured balance usage.

Kimi K3 model and price Kimi K3 API guide Kimi K3 in Codex CLI Kimi K3 vs GPT-5.6 Sol

Kimi K3 access FAQ

Can I use Kimi K3 from Russia through AnyModel?

You can call the available Kimi K3 route wherever AnyModel is reachable and your account has a supported payment method. Test from the same environment where your client will run.

What model id should I use?

Use kmc/k3 with the AnyModel OpenAI-compatible endpoint. Copy the id from the live model page if the catalog changes.

Do I receive a direct Kimi Code subscription?

No. You receive compatible API access through AnyModel. The upstream Kimi Code capacity is managed by AnyModel and is not transferred to your account.

Can I run Kimi K3 in Codex CLI?

Yes. Configure Codex with the AnyModel endpoint and set the model to kmc/k3. A dedicated guide covers the installer and manual configuration.

Test Kimi K3 on your own repository

Start with a small task, inspect the result and measured usage, then compare it with GPT-5.6 or Claude on the same prompt.

Create an AnyModel key