Claude Sonnet 4.6ChatNew
Anthropic's best everyday model — exceptional code and reasoning.
7.9sp50100%up1 rail200K ctx$0.034typical$3.36 → $16.80 /MTok
Model catalog
Every model. One endpoint.
78 chat, image, and video models behind one OpenAI-compatible key — flat per-token pricing, automatic failover across 16 providers.
No card required · $0.50 starter credit · OpenAI-compatible
Every major lab. One key.
The screening room
Straight off the network.
Real generations from the video and image rails — the prompt is the caption, the price is on the card, and every one runs on the same key as chat.
Video
6 modelsVideo · text and image to video01 / 06
Veo 2
“A lone rider crosses a mirror-flat salt lake at dawn, perfect reflection below, wide anamorphic framing, hooves echoing in the silence”
View spec
google-veo-2from $2.5/secThe chat directory
Every text model with flat per-token pricing. Latency, uptime, and volume are measured on our own traffic over the trailing 30 days — never quoted.
Uncensored AIChat
NinjaChat's uncensored model for creative and unrestricted use cases.
1.6sp50100%up1 rail32K ctx$0.005typical$0.40 → $2.50 /MTok
Gemini 3.1 ProChatNew
Google's most capable standard model — excellent creative output.
13.2sp5098.9%up1 rail1M ctx$0.025typical$2.24 → $13.44 /MTok
GPT-5Chat
OpenAI's flagship model — excellent across all task types.
8.4sp50100%up1 rail128K ctx$0.025typical$1.96 → $15.68 /MTok
Gemini 3 FlashChat
Google's ultra-fast model with a 1M token context window.
7.0sp50100%up1 rail1M ctx$0.007typical$0.65 → $3.36 /MTok
Claude Sonnet 4.5Chat
Anthropic's previous Sonnet — strong at code and analysis.
14.3sp50100%up1 rail200K ctx$0.034typical$3.36 → $16.80 /MTok
Claude Haiku 4.5Chat
Anthropic's fastest Claude variant — lightning-quick responses.
62msp50100%up1 rail200K ctx$0.011typical$1.15 → $5.60 /MTok
Claude Opus 4.6ChatNew
Anthropic's most intelligent model — use when quality is paramount.
42msp50100%up1 rail200K ctx$0.056typical$5.60 → $28.00 /MTok
Gemini 2.5 FlashChat
Gemini's high-efficiency flash variant.
1.9sp5097.2%up1 rail1M ctx$0.005typical$0.45 → $2.80 /MTok
GPT-5.4ChatNew
OpenAI's latest flagship — the most capable GPT for hard reasoning and code.
3.5sp50100%up1 rail256K ctx$0.031typical$2.80 → $16.80 /MTok
44 models · 16 providers
Chat44
GPT-5.6 LunaChat
OpenAI's high-volume GPT-5.6 model with 1.05M context and full tool support.
0.0%up1 rail1.1M ctx$0.003typical$0.35 → $1.35 /MTok
GPT-5.6 SolChat
OpenAI's flagship GPT-5.6 model for the hardest reasoning and coding workloads.
1 rail1.1M ctx$0.062typical$5.60 → $33.60 /MTok
GPT-5.6 TerraChat
OpenAI's balanced GPT-5.6 model for coding, agents, and professional work.
1 rail1.1M ctx$0.025typical$2.24 → $13.44 /MTok
GPT-OSS 120BChat
OpenAI's Apache-2.0 open-weight 120B reasoning model, served on neutral inference hosts.
809msp50100%up5 rails131K ctx$0.002typical$0.30 → $0.75 /MTok
GPT-OSS 20BChat
OpenAI's compact open-weight reasoning model on three independently operated inference rails.
193msp50100%up3 rails131K ctx$0.002typical$0.22 → $0.45 /MTok
Claude Fable 5Chat
Anthropic's highest-capability generally available model for long-running agents.
2 rails1M ctx$0.112typical$11.20 → $56.00 /MTok
Claude Opus 4.6ChatNew
Anthropic's most intelligent model — use when quality is paramount.
42msp50100%up1 rail200K ctx$0.056typical$5.60 → $28.00 /MTok
Claude Opus 5Chat
Anthropic's Opus 5 for complex agentic coding and enterprise knowledge work.
2 rails1M ctx$0.056typical$5.60 → $28.00 /MTok
Claude Sonnet 4.6ChatNew
Anthropic's best everyday model — exceptional code and reasoning.
7.9sp50100%up1 rail200K ctx$0.034typical$3.36 → $16.80 /MTok
Claude Sonnet 5Chat
Anthropic's frontier Sonnet for coding and agents, with a native 1M context window.
1.7sp50100%up2 rails1M ctx$0.034typical$3.36 → $16.80 /MTok
Gemini 2.5 ProChat
Google's pro model with a massive 1M token context.
17.3sp50100%up1 rail1M ctx$0.018typical$1.40 → $11.20 /MTok
Gemini 3 ProChat
Gemini 3 Pro — Google's latest standard-tier powerhouse.
6.5sp50100%up1 rail1M ctx$0.025typical$2.24 → $13.44 /MTok
Gemini 3.1 ProChatNew
Google's most capable standard model — excellent creative output.
13.2sp5098.9%up1 rail1M ctx$0.025typical$2.24 → $13.44 /MTok
Gemini 3.7 FlashChat
Google's newest Flash model for coding, multimodal reasoning, and high-throughput agents.
1 rail1M ctx$0.017typical$1.68 → $8.40 /MTok
DeepSeek V4 Flash 0731Chat
The most popular OpenRouter model this week, served directly through Fireworks.
10.0sp50100%up3 rails1M ctx$0.002typical$0.29 → $0.43 /MTok
DeepSeek V4 Pro 0813Chat
DeepSeek's production V4 Pro release on a no-training neutral inference host.
5.7sp50100%up3 rails1M ctx$0.013typical$1.58 → $4.75 /MTok
Kimi K2.6Chat
Moonshot's multimodal agentic model for long-horizon coding and autonomous execution.
1.2sp50100%up3 rails262K ctx$0.011typical$1.14 → $4.80 /MTok
Kimi K2.7 CodeChat
Moonshot's coding-specialized Kimi with faster, more token-efficient long-horizon execution.
19.7sp50100%up3 rails262K ctx$0.011typical$1.14 → $4.80 /MTok
Kimi K3Chat
Moonshot's Kimi K3 frontier model on Fireworks with native vision and 1M context.
3 rails1M ctx$0.036typical$3.60 → $18.00 /MTok
GLM 4.7 FlashChat
Z.ai's extremely low-cost GLM reasoning model for fast coding and tool workflows.
19.1sp50100%up1 rail203K ctx$0.002typical$0.21 → $0.55 /MTok
GLM 5.2Chat
Z.AI's GLM 5.2 agentic model served directly through Fireworks' zero-retention endpoint.
11.6sp50100%up3 rails1M ctx$0.014typical$1.68 → $5.28 /MTok
Qwen 3.6 35B A3BChat
Qwen's sparse 35B/3B-active multimodal model for efficient coding and agents.
24.6sp5080.0%up1 rail262K ctx$0.002typical$0.25 → $1.14 /MTok
Qwen 3.7 PlusChat
Qwen's fast multimodal flagship for agent loops, coding, and tool use.
11.0sp50100%up1 rail262K ctx$0.007typical$0.65 → $3.60 /MTok
Qwen 3.8 2.4T A95BChat
Qwen's open-weight 2.4T-parameter sparse-MoE flagship with 95B active parameters.
2 rails262K ctx$0.019typical$2.40 → $7.20 /MTok
Qwen 3.8 27BChat
Qwen's efficient 27B model on independently operated DeepInfra and GMI rails.
874msp50100%up2 rails1M ctx$0.007typical$0.60 → $3.84 /MTok
Qwen 3.8 MaxChat
Qwen's 2.4T-parameter sparse-MoE frontier model for autonomous long-horizon work.
1.7sp50100%up3 rails262K ctx$0.016typical$1.98 → $5.94 /MTok
Qwen3 Next 80B A3BChat
Qwen's efficient 80B/3B-active instruct model on two independent direct rails.
3.2sp50100%up2 rails262K ctx$0.003typical$0.24 → $1.32 /MTok
Mistral Medium 3.5Chat
Mistral's frontier-class multimodal model optimized for agentic coding and professional work.
1 rail256K ctx$0.018typical$1.80 → $9.00 /MTok
Mistral Small 4Chat
Mistral's efficient 119B/6.5B-active hybrid model unifying instruct, reasoning, and coding.
7.6sp50100%up1 rail256K ctx$0.002typical$0.30 → $0.75 /MTok
MiniMax M2.7Chat
MiniMax's efficient agent model for complex harnesses and multi-step productivity work.
2.7sp50100%up3 rails197K ctx$0.004typical$0.45 → $1.44 /MTok
MiniMax M3Chat
MiniMax's low-cost open-weight frontier model with native multimodality and 512K context.
2.8sp50100%up2 rails512K ctx$0.004typical$0.45 → $1.44 /MTok
Ninja AutoChat
Routes each request to an eligible text model. The resolved model's token rates determine the actual charge.
0 rails1.1M ctx—typical
InklingChat
Thinking Machines Lab's 975B multimodal open-weight generalist with 1M context.
4.0sp50100%up1 rail1M ctx$0.011typical$1.20 → $4.86 /MTok
Gemma 4 26B A4BChat
Google's sparse Gemma 4 variant with efficient multimodal reasoning on two independent hosts.
6.7sp50100%up2 rails256K ctx$0.002typical$0.28 → $0.55 /MTok
Gemma 4 31BChat
Google's popular open multimodal model with native reasoning and function calling.
1.1sp50100%up2 rails262K ctx$0.002typical$0.28 → $0.53 /MTok
HY3Chat
Tencent's reasoning and coding model on two independently operated, live-tested rails.
2.4sp50100%up2 rails262K ctx$0.002typical$0.29 → $0.73 /MTok
MiMo V2.5Chat
Xiaomi's efficient multimodal agent model, currently one of OpenRouter's most-used coding models.
6.3sp50100%up1 rail1M ctx$0.002typical$0.29 → $0.43 /MTok
MiMo V2.5 ProChat
Xiaomi's higher-capability MiMo V2.5 tier for long-context coding and tool-driven agents.
5.3sp50100%up1 rail1.1M ctx$0.004typical$0.58 → $1.04 /MTok
Muse Glimmer 30BChat
A fast 30B multimodal model with vision and function calling at budget pricing.
23.4sp50100%up1 rail131K ctx$0.004typical$0.50 → $1.80 /MTok
Nemotron 3 UltraChat
NVIDIA's 550B/55B-active frontier reasoning model for demanding agent workloads.
12.6sp50100%up2 rails262K ctx$0.007typical$0.75 → $2.88 /MTok
Nemotron 3.5 LightningChat
NVIDIA's 3B-active hybrid Mamba-Transformer reasoning model at five cents per input MTok.
5.0sp50100%up2 rails262K ctx$0.001typical$0.20 → $0.35 /MTok
Seed 2.0 CodeChat
ByteDance Seed's coding-specialized multimodal model for repository-scale engineering.
2.8sp50100%up1 rail256K ctx$0.013typical$1.20 → $7.20 /MTok
Seed 2.0 MiniChat
ByteDance Seed's low-cost multimodal reasoning model with a 256K context window.
8.1sp50100%up1 rail256K ctx$0.003typical$0.35 → $0.96 /MTok
Seed 2.0 ProChat
ByteDance Seed's full-capability multimodal reasoning model for professional agent workloads.
2.8sp50100%up1 rail256K ctx$0.013typical$1.20 → $7.20 /MTok
One base_url
Point the OpenAI SDK at NinjaChat.
The same request reaches any model in the catalog — swap one line, keep your code.
one base_url· every model
curl https://www.ninjachat.ai/api/v1/chat/completions \
-H "Authorization: Bearer $NINJACHAT_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-sonnet-4.6",
"messages": [{ "role": "user", "content": "Hello!" }]
}'FAQ
Questions people actually ask.
What is the NinjaChat API?
NinjaChat is an all-in-one AI API: OpenAI, Anthropic, Google Gemini, Fal, and other labs — chat, image, and video generation — behind one OpenAI-compatible endpoint, one prepaid key (nj_sk_), and automatic provider failover. No lock-in to a single lab.
Is the NinjaChat API compatible with the OpenAI SDK?
Yes. Point base_url at https://www.ninjachat.ai/api/v1 and use an nj_sk_ key. Streaming, tool calls, JSON modes, and models.list work. Always use the www host — the apex domain redirects and most HTTP clients strip Authorization on that redirect.
Does NinjaChat include image and video generation?
Yes. The same key calls POST /v1/chat/completions, POST /v1/images/generations (synchronous), and POST /v1/videos (submit, then poll GET /v1/videos/{id} or register a webhook). Register webhooks at https://www.ninjachat.ai/developers/keys#webhooks. Image models include FLUX, Imagen 4, Recraft, and gpt-image-2. Video models include Veo, Kling, and Seedance.
How do I switch to NinjaChat from another AI provider?
For OpenAI-compatible clients, change the base URL and API key. Step-by-step guides for OpenAI, Anthropic, OpenRouter, Replicate, fal.ai, Together AI, Fireworks AI live at https://www.ninjachat.ai/migrate. Frameworks (Vercel AI SDK, LangChain, LlamaIndex, LiteLLM) keep working.
How is the NinjaChat API priced?
Pay as you go from one prepaid balance. Chat bills metered per-token rates. Images and video bill per generation. Prices live on GET /api/v1/models — no auth required. No credit card to start; $0.50 starter credit.
Where is the OpenAPI spec and the live model catalog?
OpenAPI 3.1 at https://www.ninjachat.ai/openapi.json (also GET /api/v1/openapi). Live catalog at GET /api/v1/models (no auth). Docs at https://docs.ninjachat.ai. Status at https://www.ninjachat.ai/status. Benchmarks at https://www.ninjachat.ai/benchmarks.
One key. Every model here.
Auto-routed across providers, transparent per-token pricing, OpenAI-compatible. Start in under a minute.
Get an API key
OpenAI
Anthropic
Google
xAI
OpenRouter
Black Forest Labs
Replicate