Conol 1.5.1 — The newest models, everywhere
Conol
- release-notes
The last few weeks brought a new model from almost every provider Conol works with, and several of the models behind image, video and audio generation are being switched off by the companies that host them. This release catches up on all of it at once: the newest chat models are in the picker, the versions they replace are gone, and media generation runs on the current generation.
Ten new chat models
Why. A newer version of a model is usually better at the same job for the same price or less. You shouldn't have to wait for it, and you shouldn't have to wonder whether the one you picked is still the best in its family.
How to use it. Open the model picker when you start a chat, or switch models in the middle of one. These are new, and all ten can read images:
| Provider | Model |
|---|---|
| Anthropic | Claude Fable 5.2 |
| OpenAI | GPT-6.1 Terra |
| Gemini 4 Carbon | |
| Gemini 3.5 Flash Lite | |
| xAI | Grok 4.7 |
| Moonshot | Kimi K3.1 Pro |
| MiniMax | MiniMax M3.1 Flash |
| Qwen | Qwen 3.8 Flash |
| Tencent | Hy4 |
| StepFun | Step 5 Preview |
Two of the presets moved with them. Pro now runs on GPT-6.1 Sol, and Moderate uses Kimi K3.1 Pro when a message includes an image. If you use a preset, there's nothing to change.
A shorter list, without the older versions
Why. The picker had grown to 43 models, and many of them were earlier versions of something further up the same list. Claude Opus 4.7, 4.8, 5 and 5.5 all sat next to each other, and the GPT and Gemini families looked much the same. That makes the list harder to read and easy to pick wrongly from.
How it works. Models with a newer version are no longer offered, which leaves 30. You don't need to do anything about chats that used one of them. A chat, a timer or a saved default that points at a retired model runs on its successor from now on:
| If you were using | It now runs on |
|---|---|
| Claude Fable 5, 5.1 | Claude Fable 5.2 |
| Claude Opus 4.7, 4.8, 5 | Claude Opus 5.5 |
| Claude Sonnet 4.6 | Claude Sonnet 5.5 |
| Claude Haiku 4.5 | Claude Haiku 5.5 |
| GPT-6 Sol, GPT-5.6 Sol, GPT-5.5 | GPT-6.1 Sol |
| GPT-5.6 Terra | GPT-6.1 Terra |
| GPT-5.6 Luna | GPT-6 Luna |
| GPT-5.5 Pro | GPT-6 Astra |
| Gemini 3.1 Pro | Gemini 4 Argon |
| Gemini 3.1 Flash Lite | Gemini 3.5 Flash Lite |
| Grok 4.3 | Grok 4.7 |
| Kimi K3, Kimi K2.7 Code | Kimi K3.1 Pro |
| GLM 5.1, 5.2 | GLM 5.3 |
| Qwen 3.7 Max | Qwen 3.8 Max |
| Qwen 3.7 Plus | Qwen 3.8 Flash |
| Hy3 | Hy4 |
Tip
A successor can cost a different number of credits than the model it replaces. If you have a timer that runs often on one of the models in the left column, it's worth a look at what it uses now.
New models for images, video, music and speech
Why. The models behind media generation were a generation old, and two of them are going away: the video model the agent used by default is being removed by its host on November 17, and the image model has been deprecated by its maker. Their replacements are also simply better: sharper images, video with sound, longer music.
How to use it. Nothing changes in how you ask. Tell an agent what you want ("make a poster for the launch", "turn this photo into a ten-second clip", "read this paragraph aloud") and it picks a model. You can also name one.
| Models | Good to know | |
|---|---|---|
| Images | GPT Image 2.6 (default), FLUX 3, FLUX 3 Klein, FLUX 3 Edit, Ideogram 4.5, Recraft V4.1 | Ideogram is the one for text inside an image, such as posters and logos. FLUX 3 Edit changes an existing image from an instruction. |
| Video | Gemini Omni Flash (default), Seedance 2.5, Kling 4.0 | Clips come with sound. Omni Flash goes from 360p up to 4K and up to 15 seconds; Seedance and Kling go up to 30 seconds at up to 1080p. Give it an image to use as the first frame. |
| Music | Lyria 3.5, ElevenLabs Music v2.5, MiniMax Music 3 | You can say how long the track should be, up to five minutes with ElevenLabs and MiniMax. Ask for an instrumental if you don't want vocals. |
| Speech | Eleven v4 Turbo, Gemini 3.8 Flash TTS | Both take a voice by name. |
A few things to know:
- Cost. Video costs far more credits than an image, and the cost grows with length and resolution. A 4K clip costs about three times a 720p one of the same length. If you're trying an idea out, ask for a short clip at a low resolution first.
- Starting a video from an image. Kling 4.0 needs the image to be at least 300 × 300 pixels.
- Older model names still work. If a saved prompt or a skill names one of the previous models, the agent uses its replacement.
Fixes
- The message box on a phone — on a narrow screen, the computer picker pushed the voice and send buttons out of the message box. The model name now shortens and the computer picker shows just its icon, so every button stays in reach.
- Speech is charged by length — a text-to-speech request used to cost a flat amount. It now costs credits in proportion to the length of the text.
- Very long prompts on GPT-6 models and Grok 4.7 — these providers charge a higher rate once a single prompt passes a certain size (about 272,000 tokens for the GPT-6 models, 200,000 for Grok 4.7). Conol's rates left that out; they now include it. Ordinary chats are nowhere near that size and cost the same as before.
← Back to blog