56 2 weeks ago

vision tools thinking 309b
ollama run frob/mimo-v2.6-flash

Details

2 weeks ago

8c40795f4b68 · 178GB

mimo2
·
310B
·
(!unknown_file_type 38!)
clip
·
1.32B
·
BF16
MIT License Copyright (c) [year] [fullname] Permission is hereby granted, free of charge, to any per
{ "temperature": 1, "top_p": 0.95 }

Readme

Imported from https://hf.co/XiaomiMiMo/MiMo-V2.6-Flash-RL.

Conversion doesn’t support MTP yet.

The embedded Jinja template doesn’t handle tool calls very well, so the model needs a parser/renderer to properly process prompts and responses. Ollama doesn’t currently have a MiMo parser, so this model uses the qwen3.5 parser as a close approximation.

Chat/thinking

$ ollama run frob/mimo-v2.6-flash
>>> hello
Thinking...
Hello! I see the user just said "hello". This is a simple greeting, so I 
should respond warmly and invitingly.  

The user might be testing if I'm responsive, or perhaps they're about to 
ask something but started with a greeting. I should acknowledge their 
hello and open the door for further conversation.  

I'll keep it friendly and straightforward—no need for overcomplication. A 
smiley face would make it warmer.  

The response should be in English since the user initiated in English. 
Short and sweet feels right here.
...done thinking.

Hello! 😊 How can I help you today?

>>> /clear
Cleared session context
>>> /set nothink
Set 'nothink' mode.
>>> hello
Hello! How can I help you today?

Tools

$ ollama launch opencode \
    --model frob/mimo-v2.6-flash \
    -- \
    run 'create a file called README with the contents "hello world"'

> build · frob/mimo-v2.6-flash:latest

← Write 

Wrote file successfully.

Created `README` with "hello world" at /home/rick/ollama/issues/mimo-v2.6/README.
$ cat README
hello world

Vision

$ ollama run frob/mimo-v2.6-flash describe ./image1.jpg 
Added image './image1.jpg'
Thinking...
Here's a thinking process that leads to the description:

1.  **Analyze the Request:**
*   **Input:** An image of a white puppy sitting on a concrete surface.
*   **Task:** Describe the image.

2.  **Initial Observation & Overall Impression:**
*   The subject is a small, fluffy, white puppy.
*   It's sitting.
*   The setting appears to be outdoors or on a porch.
*   The lighting is soft, suggesting daylight but maybe slightly overcast 
or in shade.

3.  **Detailed Breakdown of Elements:**

*   **The Subject (The Puppy):**
*   **Breed/Type:** Looks like a Spitz-type or perhaps a Golden Retriever 
puppy (though less golden), or maybe a Jindo/Lab mix. It's predominantly 
white with very faint cream/tan tinting on the ears. It's definitely a 
young puppy (small, fluffy coat, slightly floppy ears).
*   **Pose:** Sitting upright on its haunches. Front legs are straight, 
supporting its weight. Body is angled slightly to the right of the frame, 
but the head is turned further to the right, looking off-camera.
*   **Features:**
*   **Fur:** Thick, plush, soft-looking white fur.
*   **Face:** Dark, round eyes (looking intently at something). A black 
nose. A slightly open mouth (maybe panting or just relaxed). Whiskers are 
visible.
*   **Ears:** Semi-floppy or button ears, sticking out slightly.
*   **Accessories:** Wearing a collar. The collar looks red with a pattern 
(maybe hearts or flowers?). It has a small, round, gold-colored bell or 
tag hanging from it.

*   **The Background/Setting:**
*   **Ground:** The puppy is sitting on a rough, grey concrete or stone 
slab/step. The edge of the step is visible in the foreground. There's some 
texture and slight discoloration on the concrete.
*   **Background (Depth):** Behind the puppy, the background is out of 
focus (shallow depth of field). There's a horizontal line suggesting a 
step or a wall (brownish/reddish tone) and a darker area above that 
(possibly a doorway or shadow). This blurs into dark brown/black at the 
very top.

4.  **Drafting the Description (Iterative Process):**

*   *Attempt 1 (Mental Outline):* There's a white puppy. It has a red 
collar with a bell. It's sitting on a concrete step. It's looking to the 
right. The background is blurry.

*   *Attempt 2 (Adding Detail):* A fluffy white puppy sits on a grey 
concrete surface. It has a red collar with a gold bell. The puppy has dark 
eyes and a black nose. It's looking away from the camera. The background 
is dark and blurry, showing maybe a brown wall or step.

*   *Attempt 3 (Refining and Polishing - similar to final output):*
*   **Subject:** A small, fluffy white puppy is the central focus.
*   **Action:** It is sitting upright on a concrete step or ledge.
*   **Gaze:** Its head is turned to the right, looking off into the 
distance with a curious or attentive expression. The eyes are dark and 
shiny.
*   **Accessories:** The puppy wears a red collar with a pattern and a 
small gold bell hanging from it.
*   **Physical traits:** The fur is thick and white, with slightly darker, 
beige-tinted ears. The nose is black.
*   **Environment:** The surface is rough grey concrete. The background is 
blurred (bokeh effect), featuring a horizontal band of reddish-brown and a 
dark upper section, likely a doorway or shadow.

5.  **Final Review:** Does the description capture the mood? Yes, it's 
cute and candid. Does it cover all main points? Yes (Subject, Action, 
Setting, Details).

6.  **Final Output Generation:** (This matches the provided good response).
...done thinking.

Based on the image, here is a description:

**Subject**
The image features a small, fluffy white puppy sitting upright. The puppy 
appears to be a young breed, possibly a Spitz or Retriever mix, with 
thick, soft-looking white fur. Its ears are slightly floppy and have a 
faint beige or cream tint compared to the rest of its body.

**Pose and Expression**
The puppy is sitting on its haunches on a concrete surface. Its body is 
angled slightly, but its head is turned sharply to the right (the viewer's 
right), looking off-camera. It has a curious, attentive expression with 
dark, glossy eyes and a black nose.

**Accessories**
The puppy is wearing a red collar with a small pattern (possibly hearts) 
and a small gold bell or round tag hanging from it.

**Setting and Background**
The puppy is perched on a rough, grey concrete step or ledge. The 
foreground shows the textured, weathered surface of the concrete. The 
background is out of focus, showing a horizontal band of reddish-brown 
(likely a wall or another step) and a dark, shadowy area at the top, which 
helps isolate the bright white puppy as the focal point.

Audio

$ ollama run frob/mimo-v2.6-flash describe ./ask.wav 
Added audio './ask.wav'
Thinking...
The user wants me to describe the audio provided.
The audio is a short clip of a man speaking.
The words are: "Ask not what your country can do for you, ask what you can 
do for your country."
This is a famous quote from John F. Kennedy's inaugural address.
The voice sounds like JFK.
The audio quality is a bit old, consistent with a recording from 1961.
There's some background noise, like a crowd or wind, typical of an outdoor 
speech.
The speaker's delivery is emphatic and clear.

So, the description should include:
- The speaker (sounds like JFK)
- The exact quote
- The tone/delivery (emphatic, clear)
- The audio quality (older recording, background noise)

Let's write it out clearly.
"The audio features a male speaker, likely John F. Kennedy, delivering a 
famous quote. He says, 'Ask not what your country can do for you, ask what 
you can do for your country.' The delivery is emphatic and clear, and the 
recording has a slightly muffled, vintage quality with some background 
noise, consistent with an older outdoor speech."
...done thinking.

The audio features a male speaker, likely John F. Kennedy, delivering a 
famous quote. He says, "Ask not what your country can do for you, ask what 
you can do for your country." The delivery is emphatic and clear, and the 
recording has a slightly muffled, vintage quality with some background 
noise, consistent with an older outdoor speech.