DR.PANDA AI PLATFORM · AI infrastructure for private deployments Explore capabilities
Dr.PandaAI PLATFORM
MODEL CATALOGUE

Choose the right
kind of intelligence.

Explore Dr.Panda model families for chat, reasoning, image, voice and multimodal applications.

TEXT & VISIONTOOLSIMAGEVOICE
DR.PANDA MODEL FAMILIES

Built around
your workload.

View API formats →

Choose a model family based on task complexity, context needs and the balance of speed and reasoning your workflow requires.

CORE MODELS↗
01

Dr.Panda Core

A versatile starting point for chat, structured tasks, visual input and tool-assisted workflows.

ChatVision inputTool calls
drpanda-coreGeneral purpose
CORE MODELS↗
02

Dr.Panda Reasoning

Designed for deeper analysis, multi-step reasoning and longer application workflows.

ChatVision inputTool calls
drpanda-reasoningGeneral purpose
MEDIA SERVICES↗
AV

Image & Voice

Generation and editing for images, with speech recognition, text processing and voice output.

GenerateEditTranscribe
media APIsMultimodal

Designed for flexible integration across language, vision, image and voice workflows.

MODEL FIT GUIDE

Match the route
to the work.

Select a model family according to the complexity, speed and context requirements of your application.

01 / EVERYDAY

Dr.Panda Core

Consider the general-purpose route for chat, extraction, short summaries and structured tasks where a balanced latency and cost profile matters.

Check: JSON reliability, multilingual quality and tool-call behavior on your own examples.

02 / DEEP WORK

Dr.Panda Reasoning

Consider a higher-capability route for multi-step analysis, planning or code review when extra reasoning makes a meaningful difference.

Check: quality lift against response time, token use and the option to fall back on simpler requests.

03 / MEDIA TASKS

Image & Voice

Use a purpose-built media path for generation or editing, transcription and speech output. Keep media operations separate from follow-up text analysis.

Check: file size, output format, audio duration, supported languages and per-operation cost.

DEPLOYMENT PATTERNS

Keep the model choice
easy to change.

04 / PRIVATE ENDPOINT

Keep credentials server-side

Route application requests through a controlled service boundary. Store keys in server configuration and avoid exposing secrets in browser bundles.

05 / ROUTING

Choose by task

Send simple, high-volume jobs to an economical route and reserve a stronger model for tasks that benefit from it. Keep routing rules observable and easy to revise.

06 / FALLBACK

Reliable model routing

Set clear timeout and retry behavior. A fallback should preserve user intent and report when it changes the model or capability.

SELECTION CHECKLIST

Three details worth
checking early.

07 / QUALITY

Evaluate real examples

Build representative prompts, expected outputs and edge cases. Review failures by type so one aggregate score does not conceal important gaps.

08 / LATENCY & COST

Measure the full call

Include routing, tool execution and media processing in estimates. Set practical limits for output length and define when a task should stop.

09 / CONTEXT

Send only what helps

Long context can help with large records, but irrelevant history adds cost and can dilute attention. Retrieve useful material and preserve concise task state.

Explore model families by capability and workflow, then pair each route with the needs of your application.

CONTEXT HANDLING

More room for
complex work.

Project records document a 262,144-token default for existing clients and negotiated support up to 327,680 tokens when explicitly requested.

DOCUMENTED CAPACITY262,144 default327,680 negotiated maximum

Extended context supports longer documents and multi-step workflows.

NEXT STEP

Connect with a
familiar API.

Explore API formats ↗