Skip to main content

AI Models & Language Support

Bike4Mind supports a wide range of state-of-the-art AI models for text generation, image generation, video generation, and speech-to-text, giving you the flexibility to choose the right model for your specific needs.

Model Availability: Model availability is managed at the organizational level by your administrator. If you don't see a model or provider listed below in your model selector, it may be disabled for your organization or restricted to certain user roles. Contact your organization administrator to request access to additional models or providers.

Supported Text Models​

OpenAI Models​

GPT-5 Family​

  • GPT-5.5 — Latest and most capable flagship, ~1M context window (1.05M tokens)
  • GPT-5.4 — Highly capable previous flagship, 1M+ context window
  • GPT-5.4 Mini — Compact GPT-5.4 variant, 400K context window
  • GPT-5.4 Nano — Ultra-lightweight GPT-5.4, 400K context window
  • GPT-5.2 — Previous-generation flagship, 400K context window
  • GPT-5.2 Chat Latest — Always points to the latest GPT-5.2 version
  • GPT-5.1 — Early GPT-5 series, 400K context window
  • GPT-5.1 Chat Latest — Always points to the latest GPT-5.1 version
  • GPT-5 — Original GPT-5 release, 400K context window
  • GPT-5 Mini — Compact version with excellent performance at lower cost
  • GPT-5 Nano — Ultra-efficient variant for lightweight tasks
  • GPT-5 Chat Latest — Always points to the latest GPT-5 version

GPT-4 Family​

  • GPT-4.1 — Enhanced GPT-4 with 1M+ token context window
  • GPT-4.1 Mini — Cost-effective variant with the same 1M+ context window
  • GPT-4.1 Nano — Ultra-efficient variant with the same 1M+ context window
  • GPT-4.5 Preview — Preview model bridging GPT-4 and GPT-5
  • GPT-4o — Multimodal model with 128K context
  • GPT-4o Mini — Cost-effective multimodal model
  • GPT-4 Turbo — High-performance model with 128K context
  • GPT-4 — Original GPT-4 model

O-Series (Reasoning Models)​

  • O4 Mini — Latest compact reasoning model with agentic capabilities
  • O3 — Advanced reasoning with structured thinking
  • O3 Mini — Smaller, faster reasoning model
  • O1 — Original reasoning model with 200K context
  • O1 Preview — Early reasoning model preview
  • O1 Mini — Compact reasoning model

Anthropic Models (Claude)​

Anthropic models are available via both the Anthropic API and AWS Bedrock. Your administrator determines which backend is active for your organization.

Claude Fable Series​

  • Claude Fable 5 — Anthropic's most capable model, built for complex long-running agentic tasks, with 1M context and adaptive reasoning. Generally available and selectable in the model picker. Its safety classifiers may decline some requests (targeting cybersecurity and research-biology content); when that happens the request transparently continues on Claude 4.8 Opus rather than failing.

Claude 4.8 Series​

  • Claude 4.8 Opus — Latest generally available Claude flagship, 1M context window with adaptive reasoning

Claude 4.7 Series​

  • Claude 4.7 Opus — High-capability reasoning and analysis, 1M context window with adaptive reasoning

Claude 4.6 Series​

  • Claude 4.6 Opus — Powerful reasoning and analysis, 1M context window
  • Claude 4.6 Sonnet — Enhanced coding, analysis, and complex reasoning with improved speed and efficiency

Claude 4.5 Series​

  • Claude 4.5 Opus — Top-tier reasoning and analysis
  • Claude 4.5 Sonnet — Balanced performance and cost
  • Claude 4.5 Haiku — Fast and lightweight with large output capacity

Claude 4 Series​

  • Claude 4.1 Opus — Enhanced Claude 4 with improved performance
  • Claude 4 Opus — Powerful reasoning and analysis
  • Claude 4 Sonnet — Balanced model with competitive pricing

Claude 3 Series​

  • Claude 3.7 Sonnet — Most advanced Claude 3.x model with thinking support
  • Claude 3.5 Sonnet / Claude 3.5 Sonnet V2 — Balanced performance and cost
  • Claude 3.5 Haiku — Fast and lightweight
  • Claude 3 Opus — Original top-tier Claude model

Claude 3.x through 4.5 models (and Claude 4.6 Sonnet) support a 200K token context window. Claude 4.6 Opus and the newer Opus models (4.7, 4.8) and Claude Fable 5 support a 1M token context window.

Google Gemini Models​

Gemini 3.5​

  • Gemini 3.5 Flash — Latest generation, fast and efficient with 1M context window and thinking support

Gemini 3.1 (Preview)​

  • Gemini 3.1 Pro Preview — Advanced reasoning with 1M context window and thinking support
  • Gemini 3.1 Flash Lite — Ultra-lightweight latest-gen variant with 1M context window and thinking support

Gemini 3 (Preview)​

  • Gemini 3 Pro Preview — Earlier Gemini 3 generation, 1M context window
  • Gemini 3 Flash Preview — Speed-optimized Gemini 3 generation

Gemini 2.5​

  • Gemini 2.5 Pro — Flagship model with 2M context window and thinking support
  • Gemini 2.5 Flash — Fast and efficient with 1M context window and thinking support
  • Gemini 2.5 Flash Lite — Ultra-lightweight variant

Gemini Legacy​

  • Gemini 2.0 Flash Experimental — Experimental next-gen model
  • Gemini 1.5 Pro — 2M context window for massive documents
  • Gemini 1.5 Flash — Speed-optimized with 1M context
  • Gemini 1.5 Flash 8B — Compact variant

xAI Models (Grok)​

  • Grok 4 — Latest generation with 256K context window
  • Grok 3 — Powerful reasoning model with 131K context
  • Grok 3 Fast — Speed-optimized Grok 3
  • Grok 3 Mini — Compact Grok 3
  • Grok 3 Mini Fast — Fastest Grok variant
  • Grok 2 — Previous generation
  • Grok 2 Vision — Multimodal with image understanding

Meta Llama Models (AWS Bedrock)​

  • Llama 4 Maverick 17B Instruct — Optimized for fast, efficient inference
  • Llama 4 Scout 17B Instruct — Specialized for exploration and reasoning
  • Llama 3 Instruct 70B — Powerful open-source model
  • Llama 3 Instruct 8B — Compact model for high-volume tasks

DeepSeek Models (AWS Bedrock)​

  • DeepSeek R1 — Advanced reasoning model with 128K context
  • DeepSeek v3.1 — General-purpose model with 128K context

Amazon Titan Models (AWS Bedrock)​

  • Titan Text Express — General-purpose text generation
  • Titan Text Lite — Lightweight variant

AI21 Labs Models (AWS Bedrock)​

  • Jurassic-2 Ultra — High-capability text generation
  • Jurassic-2 Mid — Balanced performance

Local Models (Ollama)​

If your organization has Ollama configured, you can access locally-hosted models including:

  • LLaMA 3.3 — Meta's latest local model
  • DeepSeek R1 — Local reasoning model
  • TinyLlama — Ultra-lightweight local model
  • Any other models installed on your organization's Ollama instance

Supported Image Models​

FLUX Models (Black Forest Labs)​

  • FLUX Pro Ultra — Highest quality, photorealistic output
  • FLUX Pro 1.1 — Professional quality generation
  • FLUX Pro (Legacy) — Previous generation
  • FLUX Pro Fill — Inpainting and fill operations
  • FLUX Kontext Pro — Image-to-image transformations
  • FLUX Kontext Max — Maximum quality transformations

OpenAI Image Models​

  • GPT-Image-1.5 — Latest generation image model
  • GPT-Image-1 — Previous generation
  • GPT-Image-1 Mini — Cost-effective variant

Google Gemini Image Models​

  • Gemini 3 Pro Image Preview — Google's latest image generation
  • Gemini 2.5 Flash Image — Fast image generation

xAI Image Models​

  • Grok 2 Image — Image generation via xAI

Supported Video Models​

OpenAI Sora​

  • Sora — AI video generation (4s, 8s, or 12s clips)
  • Sora Pro — Higher quality video generation

Speech-to-Text Models​

  • Whisper-1 (OpenAI) — Industry-leading speech recognition
  • Amazon Transcribe (AWS) — Cloud-based transcription service

Model Configuration​

Temperature (0.0 - 2.0)​

Controls randomness in responses:

  • 0.0: Most deterministic, factual
  • 0.7: Balanced creativity
  • 1.0: Default setting
  • 2.0: Maximum creativity

Max Tokens​

Controls response length:

  • Automatically adjusted based on model capabilities
  • Context-aware limits to prevent overflow
  • Model-specific maximums respected

Advanced Settings​

Streaming​

  • Real-time token-by-token generation
  • Reduces perceived latency
  • Available for most models

Multiple Responses (n)​

  • Generate 1-4 parallel responses
  • Compare different approaches
  • Useful for creative tasks

Thinking Models​

  • Explicit reasoning steps before responding
  • Configurable token budget
  • Available for models flagged with thinking support — including O4 Mini, the GPT-5.4 family, Claude 3.7 Sonnet and all Claude 4.x models (Sonnet, Haiku, Opus, Fable), and Gemini 2.5 Pro/Flash plus all Gemini 3.x text models

Model Selection Best Practices​

For General Use​

  • GPT-4o Mini or Claude 4.5 Haiku: Best balance of cost and performance
  • Gemini 2.5 Flash: Fast responses for simple queries

For Complex Tasks​

  • Claude 4.8 Opus or GPT-5.5: Deep analysis and reasoning
  • O3 or O4 Mini: Mathematical and logical problems
  • Gemini 2.5 Pro: Massive document analysis (2M context)

For Creative Writing​

  • Claude 4.5 Sonnet: Nuanced creative output
  • GPT-5: Large context for long-form content
  • Temperature: 0.8-1.2

For Code Generation​

  • Claude 4.5 Sonnet or GPT-4.1: Excellent code comprehension
  • DeepSeek R1: Strong reasoning for complex code tasks
  • Temperature: 0.3-0.5

For Image Generation​

  • FLUX Pro Ultra: Photorealistic images
  • GPT-Image-1.5: Creative and artistic
  • FLUX Kontext Pro/Max: Image transformations and editing

For Video Generation​

  • Sora Pro: Highest quality AI video clips

Context Windows​

Different models support different context lengths:

ModelContext WindowBest For
GPT-4.1 / 4.1 Mini / 4.1 Nano1M+ tokensMassive codebases
GPT-5.5~1M tokensLong documents, latest flagship
GPT-5.4~1M tokensLong documents, deep analysis
GPT-5 / 5.1 / 5.2 (incl. GPT-5 Mini/Nano), GPT-5.4 Mini/Nano400K tokensLong documents
Claude 4.6 Opus / Claude 4.7 Opus / Claude 4.8 Opus / Claude Fable 51M tokensLargest Claude context
Claude (3.x–4.5, 4.6 Sonnet)200K tokensExtensive analysis
Gemini 2.5 Pro2M tokensLargest documents
Gemini 3.x / 2.5 Flash1M tokensLarge documents, fast
Grok 4256K tokensLong conversations
DeepSeek R1 / v3.1128K tokensCode repositories

Cost Optimization​

Tips for Managing Credits​

  1. Start with smaller models for initial drafts (Haiku, Mini, Flash variants)
  2. Use streaming to stop generation early if needed
  3. Set appropriate max tokens to avoid waste
  4. Choose the right model tier for the task complexity

Model Cost Tiers​

  • Economy: GPT-4o Mini, Claude 4.5 Haiku, Gemini Flash, Grok 3 Mini Fast
  • Standard: GPT-4.1, Claude 4.5 Sonnet, Gemini 2.5 Flash, Grok 3
  • Premium: GPT-5, Claude 4.5 Opus, Gemini 2.5 Pro, Grok 4
  • Ultra: O3, Claude 4.6 Opus / Claude 4.7 Opus / Claude 4.8 Opus, GPT-5.5

API Keys & Custom Models​

Bring Your Own Keys​

  • Use your own API keys for supported providers (OpenAI, Anthropic, Google, xAI, Black Forest Labs)
  • Manage keys from your Profile settings

Ollama Integration​

  • Connect locally-hosted models
  • Custom model support
  • Privacy-first approach

Need a Model Enabled?​

If a model or provider you need isn't available in your model selector, contact your organization administrator. They can:

  • Enable or disable specific models for your organization
  • Configure which providers are active
  • Adjust model access based on user roles