NVIDIA-hosted Kimi K2.5 / K2.6: free chat + API key, no payment details
Source: https://build.nvidia.com/moonshotai/kimi-k2.5/modelcard
Description
Create account to comment on specific lines or Sign in
+
1
Moonshot's own platform at platform.moonshot.ai requires a $1 minimum top-up before you can issue an API key — there is no truly free tier. NVIDIA solves that for you: it hosts Moonshot's flagship Kimi K2.5 (and the newer Kimi K2.6) on its build.nvidia.com inference catalogue, where any free NVIDIA Developer Program account gets a playground for in-browser chat plus an OpenAI-compatible API key (prefix nvapi-) without entering a credit card. Rate-limited (commonly ~40 requests/min per model) but otherwise unmetered for prototyping — a clean way to evaluate Moonshot's Kimi line without paying Moonshot a cent.
No comments on this line yet.
+ 2
No comments on this line yet.
+
3
No comments on this line yet.
+ 4
No comments on this line yet.
+ 6
No comments on this line yet.
+ 7 1. Go to the model card: build.nvidia.com/moonshotai/kimi-k2.5 (or /kimi-k2.6 for the newer revision).
No comments on this line yet.
+ 8 2. Click "Try API" / "Get API Key" in the right panel.
No comments on this line yet.
+ 9 3. Sign in or create a free NVIDIA account — email + password, or SSO (Google/GitHub-style federated). No credit card, no phone SMS.
No comments on this line yet.
+ 10 4. Confirm your email — this enrolls you in the NVIDIA Developer Program automatically.
No comments on this line yet.
+
11
5. Back on the model card, click "Get API Key" again → it generates an nvapi-... token. Copy it immediately — it is shown once.
No comments on this line yet.
+ 12 6. Use it as a drop-in OpenAI key:
No comments on this line yet.
+ 13
No comments on this line yet.
+
14
No comments on this line yet.
+ 15 from openai import OpenAI
No comments on this line yet.
+ 16 client = OpenAI(
No comments on this line yet.
+
17
base_url="https://integrate.api.nvidia.com/v1",
base_url="https://integrate.api.nvidia.com/v1", No comments on this line yet.
+
18
api_key="nvapi-...",
api_key="nvapi-...", No comments on this line yet.
+ 19 )
No comments on this line yet.
+ 20 resp = client.chat.completions.create(
No comments on this line yet.
+
21
model="moonshotai/kimi-k2.5",
model="moonshotai/kimi-k2.5", No comments on this line yet.
+
22
messages=[{"role": "user", "content": "hello"}],
messages=[{"role": "user", "content": "hello"}], No comments on this line yet.
+ 23 )
No comments on this line yet.
+
24
No comments on this line yet.
+ 25
No comments on this line yet.
+ 26 Important:
No comments on this line yet.
+ 27 • The same key works for all 100+ models in the NIM catalogue (Llama Nemotron, GLM, MiniMax, DeepSeek, etc.) — not just Kimi.
No comments on this line yet.
+ 28 • The legacy "1,000 credits on signup" scheme was retired in early 2025; today it's purely rate-limited, no credit balance to exhaust.
No comments on this line yet.
+ 29 • Some accounts get a "Request More" button visible on build.nvidia.com that bumps the rate cap (historically up to ~5,000 inference units), useful if you hit limits during a hackathon.
No comments on this line yet.
+ 30 • Account verification is occasionally manual for new sign-ups — if your key has zero quota, post in the NVIDIA Developer Forums (Access/Accounts) for a fix.
No comments on this line yet.
+ 31
No comments on this line yet.
+
32
No comments on this line yet.
+ 33
No comments on this line yet.
+ 35
No comments on this line yet.
+ 36 ModelEndpoint slugNotes
No comments on this line yet.
+
37
Kimi K2.5moonshotai/kimi-k2.5Trillion-parameter MoE, ~32B active. Multimodal (text + vision via the kimi-k2.5 VLM card). 256K context.
No comments on this line yet.
+
38
Kimi K2.5 Multimodal VLMmoonshotai/kimi-k2.5 (vision branch)Image + text input, GPU-accelerated NIM endpoint.
No comments on this line yet.
+
39
Kimi K2.6moonshotai/kimi-k2.6Newer revision — improved reasoning + tool use. Same free-tier policy.
No comments on this line yet.
+ 40
No comments on this line yet.
+
41
All endpoints are OpenAI Chat Completions-compatible at https://integrate.api.nvidia.com/v1 — works with the official openai SDK in Python/Node, plus drop-in tools like Kilo Code, OpenCode, Cline, Continue, and aider. See the official model list.
No comments on this line yet.
+ 42
No comments on this line yet.
+
43
No comments on this line yet.
+ 44
No comments on this line yet.
+ 46
No comments on this line yet.
+ 47 LimitValue
No comments on this line yet.
+ 48 Cost$0 — no credit card, no auto-billing
No comments on this line yet.
+ 49 Rate limit~40 requests/minute (varies by model and current cluster load)
No comments on this line yet.
+ 50 Token quotaNone disclosed — effectively unmetered while staying under RPM
No comments on this line yet.
+ 51 ConcurrencySoft limits; long context requests can return 429 under load
No comments on this line yet.
+ 52 Trial expiryNone — Developer Program membership doesn't expire
No comments on this line yet.
+ 53 Production useAllowed for development, testing, prototyping, research. For production traffic, NVIDIA expects you to either self-host the NIM container (free 90-day NVIDIA AI Enterprise trial) or move to a paid hosted partner.
No comments on this line yet.
+ 54
No comments on this line yet.
+
55
No comments on this line yet.
+ 56
No comments on this line yet.
+ 58
No comments on this line yet.
+ 59 NVIDIA NIM (free)platform.moonshot.ai
No comments on this line yet.
+ 60 Credit card / paymentNot required$1 minimum top-up required to activate
No comments on this line yet.
+ 61 Free creditsNone disclosed, but unmetered under 40 RPM$5 voucher unlocks after $5 cumulative recharge
No comments on this line yet.
+
62
Endpointintegrate.api.nvidia.com/v1api.moonshot.ai/v1
No comments on this line yet.
+ 63 Model freshnessK2.5 + K2.6 currently mirroredAlways-newest models land here first
No comments on this line yet.
+ 64 Rate scalingFixed RPMScales with cumulative recharge
No comments on this line yet.
+ 65 Best forFree evaluation, prototyping, fallback APIProduction once you've validated and need higher throughput
No comments on this line yet.
+ 66
No comments on this line yet.
+ 67 Bottom line: if you just want to run Kimi against your prompts/agents without committing money, NVIDIA's hosted NIM is the path. If you're already shipping and need scale, eventually move to Moonshot's own platform (or OpenRouter at ~$0.44/M input, ~$2/M output for Kimi K2.5).
No comments on this line yet.
+ 68
No comments on this line yet.
+
69
No comments on this line yet.
+ 70
No comments on this line yet.
+ 72
No comments on this line yet.
+
73
• Region: build.nvidia.com is globally accessible, but the NVIDIA Developer Program signup may show extra KYC fields for some sanctioned regions. EU/US/UK/Asia-Pacific are smooth.
No comments on this line yet.
+ 74 • NVIDIA may rate-limit aggressively on popular models during launch periods or peak hours — expect 429s on Kimi K2.6 in the days right after release.
No comments on this line yet.
+ 75 • Same key, shared bucket: if you blast Llama Nemotron at 40 RPM, you may temporarily reduce headroom on Kimi calls.
No comments on this line yet.
+ 76 • No SLA, no support tickets on the free tier — for outages, watch the NVIDIA Developer Forums.
No comments on this line yet.
+ 77 • Output tokens are capped per request (typically 4K-8K depending on model) — for long generations, chunk and stream.
No comments on this line yet.
+ 78 • No fine-tuning, no batch API on the free tier — chat completions and embeddings only.
No comments on this line yet.
+ 79
No comments on this line yet.
+
80
No comments on this line yet.
+ 81
No comments on this line yet.
+ 83
No comments on this line yet.
+ 84 • OpenRouter also exposes Kimi K2.5 with no upfront recharge required (pay-as-you-go from any topped-up OpenRouter balance) — good plan-B if NIM rate-limits you.
No comments on this line yet.
+ 85 • Z.AI's GLM-4.5-Flash / 4.7-Flash / 4.6V-Flash are unconditionally free via API and broadly comparable to Kimi for many tasks — see the separate Z.AI entry in this dataset.
No comments on this line yet.
+
86
• Pair NIM-hosted Kimi with coding agents (Kilo Code, OpenCode, Cline, aider, Continue) by setting OPENAI_BASE_URL=https://integrate.api.nvidia.com/v1 and OPENAI_API_KEY=nvapi-....
No comments on this line yet.
+ 87 • The NVIDIA Developer Program also unlocks free credits/discounts on NVIDIA Brev GPU instances and DGX Cloud Lepton trials — same login.
No comments on this line yet.
+ 88
No comments on this line yet.
+
89
No comments on this line yet.
+ 90
No comments on this line yet.
+ 91 Sources:
No comments on this line yet.
+ 92 • Kimi K2.5 model card on build.nvidia.com
No comments on this line yet.
+ 93 • Kimi K2.6 NIM API reference
No comments on this line yet.
+ 94 • Build with Kimi K2.5 Multimodal VLM Using NVIDIA GPU-Accelerated Endpoints (NVIDIA Technical Blog)
No comments on this line yet.
+ 95 • Access to NVIDIA NIM Now Available Free to Developer Program Members (NVIDIA Technical Blog)
No comments on this line yet.
+ 96 • NVIDIA NIM API Explained: Free AI Inference in 2026
No comments on this line yet.
+ 97 • How to Use NVIDIA NIM + Kilo Code to Access Kimi K2.5 (Kilo blog)
No comments on this line yet.
+ 98 • Free Kimi K2.5 API via NVIDIA NIM: OpenClaw Fallback Setup
No comments on this line yet.
+ 99 • Kimi K2.5 Free API guide (Medium / Mehul Gupta)
No comments on this line yet.
+ 100 • Moonshot platform recharge & rate limits
No comments on this line yet.
+ 101 • Kimi K2.5 Pricing 2026 (NxCode)
No comments on this line yet.
+ 102 • "Request More" (+4,000 credits) option on build.nvidia.com (NVIDIA Developer Forums)
No comments on this line yet.