Two new tools and a long list of models arrived in the last ten days. Here is the short version.
A docs assistant in your dashboard
Your dashboard now has an Assistant page. It is a chat that knows the GetVibe documentation: ask how to connect Claude Code, why a request returned 429, how to set reasoning effort or which endpoint a tool needs, and it answers with the exact settings. It runs on a fast, inexpensive chat model and is billed per token from your balance, like any API call, so a typical question costs a fraction of a cent.
A public price calculator
The price calculator answers "what will this cost me per month?" before you write a line of code. Pick a model and a channel, enter your monthly input and output tokens and the share of input you expect to hit the cache, and it shows your monthly cost next to the same volume at the model maker's official price. It reads the same live prices as the models page.
New text models
All of them work with the same key and base URL, through /v1/chat/completions. Live prices per 1M tokens:
- MiniMax M3.1 Flash (
minimax-m3.1-flash) — 1M-token context and very long answers. $0.0129 input · $0.0515 output per 1M tokens - Kimi K2.8 (
kimi-k2.8) — a preview release of Moonshot's newest Kimi, with image input. $0.038 input · $0.16 output per 1M tokens - Qwen3.8 2.4T-A95B (
qwen3.8-2.4t-a95b) — the large Qwen3.8 mixture-of-experts model, text only. $0.02 input · $0.06 output per 1M tokens - Qwen3.8 27B (
qwen3.8-27b) — a compact Qwen3.8 for high-volume work. $0.005 input · $0.03 output per 1M tokens - MiMo v2.5, v2.5 Pro, v2.6 Flash and v2.6 Pro (
mimo-v2.6-proand siblings) — Xiaomi's MiMo family. $0.3108 input · $0.6215 output per 1M tokens for v2.6 Pro. - Doubao Seed 2.0 Mini, 2.0 Lite, 2.0 Code and 2.1 Turbo (
doubao-seed-2.1-turboand siblings) — ByteDance's Seed models, served from ByteDance's own endpoint. $0.7536 input · $3.7679 output per 1M tokens for 2.1 Turbo. - DeepSeek V4.1 Flash 1017 (
deepseek-v4.1-flash-1017) — the October 17 build of DeepSeek V4.1 Flash with fresher knowledge. $0.0108 input · $0.0429 output per 1M tokens
Before announcing them we sent a live request to each one through getvibeapi.com, and every one answered.
New image and video models
- Hunyuan Image 3.5 (
hunyuan-image-3-5) — Tencent's text-to-image model, with up to 5 reference images, at 1K, 2K or 4K. - Kling 3.0 Omni Reference and Transformation — video from up to 7 reference images or a reference video, and restyling of an existing clip from a text description. Details and examples in our Kling 3.0 Omni guide.
- P-Video-2 Pro Reference — talking video from photos and a voice track, which can keep your audio as the clip's sound. See the P-Video-2 Pro Reference guide.
Media models are called through the media generation API and priced per image or per second of video on the models page.
Try it
curl https://getvibeapi.com/v1/chat/completions \
-H "Authorization: Bearer $GETVIBE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "minimax-m3.1-flash",
"messages": [{"role": "user", "content": "Summarize this changelog in three bullets: ..."}]
}'Everything above is live now in your getvibeapi.com dashboard.