Not runnable todayVideo and image Q&AVerified 2026-11-05
InternVideo2.5 Chat 8B
OpenGVLab/InternVideo2_5_Chat_8B
InternVideo2.5 Chat 8B is a video and text video understanding model (Apache-2.0). Not runnable on Mixpeek today.
Something blocks it today, such as the license, remote code or a GPU-only runtime. The reason is below.
No demo for InternVideo2.5 Chat 8B yet
A video, image, slideshow or playground goes here once we record one on our own data.
What it is good and bad at
- Best at
- Long-video question answering and fine-grained temporal grounding with an open model (long and rich context modeling).
- Worst at
- 8.4B parameters and remote code: GPU only, slow per clip; answers need checking against transcripts.
- Task
- Video and image Q&A
- Input
- Video, Text
- Output
- Text
- Use cases
- Video understanding, Captioning
Templates that use it
No template uses this model yet. Browse templates.
Input and output
Input
- video
- frames
- up to hundreds of frames (dynamic sampling)
- resolution
- 448x448 tiles
- prompts
- free-form question
- max_tokens
- LLM context per the card
Output
- text
- free-form answer or JSON when prompted
How Mixpeek wires it
- Extractors
- None today
- Feature URIs
- None today
- Vector indexes
- None
- Retriever stages
- None
- Combines with
- scene splits -> InternVideo2.5 descriptions and tags -> E5 text vectors + BM25
How it runs
- Status
- Not runnable today Requires trust_remote_code (custom_code on the hub) and a 24 GB class GPU. Would run as a GPU batch plugin after vendoring the InternVL chat code.
- Serve apps
- None
- Runtime
- transformers AutoModel with trust_remote_code=True (InternVL chat)
- Weights
- safetensors, needs remote code
- Parameters
- 8,401 M
- Size on disk
- 16,802 MB
- Preprocessing
- Sample frames, tile to 448, normalize; build the chat prompt with frame tokens.
- Postprocessing
- Decode text; parse JSON when requested.
- Serving
- batch
- Hardware
- Needs a GPU
Scale to zero
- Minimum replicas
- 0
- Maximum replicas
- 1
- Idle timeout
- 600 s
- Memory
- 32 GB
- CPUs
- 8
- GPUs
- 1
- Cold start, node warm
- Not recorded
- Cold start, new node
- Not recorded
Measured cost
Unmeasured
No committed measurement; plan row 3 measures it in the standalone image.
Our evaluations
Unmeasured
We have not evaluated this model on our own data yet. Publisher results are in the sources below.
Questions
- What is InternVideo2.5 Chat 8B best at?
- Long-video question answering and fine-grained temporal grounding with an open model (long and rich context modeling).
- What are the limits of InternVideo2.5 Chat 8B?
- 8.4B parameters and remote code: GPU only, slow per clip; answers need checking against transcripts.
- Can InternVideo2.5 Chat 8B be used commercially?
- Yes. Apache-2.0 per the hub tag; the language model inside (InternLM/Qwen based InternVL) carries its own terms, check before shipping.
- How does InternVideo2.5 Chat 8B run on Mixpeek?
- Requires trust_remote_code (custom_code on the hub) and a 24 GB class GPU. Would run as a GPU batch plugin after vendoring the InternVL chat code.