Frontier AI access
for everything you build
Claude, OpenAI, Gemini, code, images, and long-context work. Native protocols for builders.
Agent
projects/mainrunning
› codex5 codex "wire Claude Code"
◆ Reading ~/.claude/settings.json [done]
◆ Thinking
Need: preserve SDK calls, swap endpoint, attach project usage.
◆ Plan
1. keep Anthropic client surface
2. route through Codex5 /v1
3. stream logs into console
◆ Edit config/ai-client.ts
12 import { Anthropic } from "@anthropic-ai/sdk";
18 baseURL: "https://api.anthropic.com",
18 baseURL: "https://www.codex5.net/v1",
19 apiKey: process.env.CODEX5_API_KEY,
20 headers: { "x-rb-project": projectId },
› Streaming model routes, retries, and usage logs
◆ Run npm test -- ai-client
✓ 4 tests passed in 1.2s
ready Claude Code now runs through Codex5.
Build
Personalize your workspace
One image. A whole new Codex.
Codex5 Pro+ turns one image into a complete Codex theme. Upload a picture, adjust the mood, and apply a workspace that feels like yours.
Upload one image
PNG, JPG, and WebP images are supported.
Local image processing
Your image stays on the current computer.
Built-in Simplified Chinese
Switch fixed interface text without changing your work.
Apply and keep it
Your theme remains after reopening Codex.
联系我们
微信客服与 QQ 社区
接入帮助、充值、密钥配置和服务通知，都可以通过微信客服或 QQ 群联系。
0.1x当前服务倍率
群号：263489143
New workspace
Image, video, and Agent flows on one canvas
Codex5 Canvas turns prompt exploration into a visual workflow. Create API keys in the main console, then compose image generation, video tasks, prompt cards, and Agent help on an infinite board.
Image generationCreate, edit, and reuse visual references.
Video tasksConnect images, video, and audio material.
Agent supportRefine prompts and plan node workflows.
Cloud syncKeep boards available across devices.
Generated in the new workspace
These featured results are created from reusable prompts and gpt-image-2 workflows in Codex5 Canvas.
Rain-window Portrait
Neon reflections, shallow depth of field, and restrained cinematic mood.
Vintage Barbershop Portrait
Classic film color, blinds light, mirrors, leather chair, and grounded character.
Coffee Bean Still Life
Warm side light, kraft packaging, ceramic cup, and commercial aroma detail.
Rain Courtyard Teahouse
Quiet timber room, rainy garden, stone lantern, moss, bamboo, and warm interior.
Converted Factory Restaurant
Red brick, steel beams, high windows, plants, wood tables, and warm dining light.
Mushroom Market
A rich forest market with wet watercolor texture and tiny fantasy silhouettes.
Forest Temple Line Art
Detailed roots, stairs, vines, stone symbols, and a clean coloring-book structure.
Mini Mars Base
A believable toy-scale Mars habitat with white pods, rovers, and red rock.
AI Workflow Blocks
Modular blocks, arrows, glowing nodes, and a readable no-text process map.
Sponge City System
Isometric rain gardens, permeable paving, storage, and water flow layers.
Eastern Tea Still Life
Porcelain, pastries, linen, soft side light, and a restrained tabletop rhythm.
Chocolate Atelier
Tempered chocolate, marble counter, copper tools, cocoa dust, and craft detail.
Cyber Rain Alley
Rain reflections, convenience-store glow, cables, street steam, and no signage.
Ancient Star Observatory
Stone instruments, deep blue skies, distant mountain lights, and ritual calm.
Seasonal Fine-dining Visual
Dark stone, precise plating, seasonal botanicals, and refined menu-cover mood.
For developers
Image generation
Image generation
Native editing
apihttps://www.codex5.net/v1
imagehttps://www.codex5.net/v1
Usagemedia group billed by usage
POST /v1/images/generations
POST /v1/images/edits
image-2 route
POST https://www.codex5.net/v1/images/generations
curl "https://www.codex5.net/v1/images/generations" \
-H "Authorization: Bearer $CODEX5_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-image-2",
"prompt": "A clean product render of an API console on a glass desk",
"size": "1024x1024",
"n": 1
}' | jq -r '.data[0].b64_json' | base64 -d > out.pngPOST https://www.codex5.net/v1/images/edits
curl "https://www.codex5.net/v1/images/edits" \
-H "Authorization: Bearer $CODEX5_API_KEY" \
-F "model=gpt-image-2" \
-F "image[]=source.png" \
-F "prompt=Change the background to Tokyo at night" \
-F "size=1024x1024" | jq -r '.data[0].b64_json' | base64 -d > out.png
sample output1.2s
promptok
renderok
costusage-based
Latest
Recent updates
Claude Fable 5 and Claude Sonnet 5 are live
Claude Fable 5 is suited to complex reasoning, long-running agents, coding, and research. Claude Sonnet 5 focuses on fast reasoning for coding, analysis, tool use, and long-context workflows.
Codex5 Canvas available
A new infinite canvas for image generation, video tasks, prompt libraries, and Agent-assisted creative workflows. Create a key in the main console and start from canvas.codex5.com.
Gemini via gemini-lite
Use the gemini-lite group for Gemini access, currently announced at 0.8 account balance per $1 official usage.
Claude Opus 4.8 available
New flagship route for complex reasoning, coding, long-context review, and tool-heavy workflows.
Choose how to get started
Build on your own
For individual builders and small teams that want direct API access.
- Access Claude, OpenAI, and Gemini model groups
- Usage-based billing with shared balance
- Compatible with existing SDKs and CLI tools
- Docs for Claude Code, Codex, Cursor, and more
Get extra support
For teams that need onboarding, invoices, higher limits, or dedicated routing.
- Dedicated setup support
- Custom rate limits and route policies
- Usage review and monthly billing options
- DPA, SLA, and isolated deployment discussion