◆Franklin/Docs

Models

60+ models from 12+ providers, including a free tier.

Available Providers

Franklin gives you access to 60+ models from 12+ providers through a single interface. No API keys needed — everything is routed through BlockRun's unified gateway. You pay provider cost + 5%, settled instantly in USDC via x402.

  • Anthropic — Claude Opus 4.8 (1M context), Claude Opus 4.7 (1M context), Claude Opus 4.5, Claude Sonnet 4.6, Claude Haiku 4.5
  • OpenAI — GPT-5.5, GPT-5.4, GPT-5.4 Pro, GPT-5.3 Codex, GPT-5.3, GPT-5.2, GPT-5.4 Mini, o1, o3
  • Google — Gemini 3.1 Pro, Gemini 3 Pro Preview, Gemini 3.5 Flash, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash
  • DeepSeek — DeepSeek V4 Pro (1M context), DeepSeek V4 Flash (chat + reasoner)
  • Moonshot — Kimi K2.6 (262K context, vision + reasoning)
  • Z.AI — GLM-5.1, GLM-5, GLM-5 Turbo
  • MiniMax — MiniMax M2.7 (204K context, reasoning)
  • xAI — Grok Imagine (image + video)
  • NVIDIA — DeepSeek V4 Flash, Nemotron, Llama 4, Qwen3, Mistral Small 4 (all free)

List All Models

See the full list of available models and their current status:

bash
franklin models

The output shows each model's name, provider, tier, and cost per million tokens.

Free Tier

The following models are always free — no wallet, no USDC, no sign-up:

  • DeepSeek V4 Flash (via NVIDIA, 1M context)
  • Nemotron 3 Nano Omni (via NVIDIA, vision)
  • Qwen3 Coder 480B & Qwen3-Next 80B (via NVIDIA)
  • Llama 4 Maverick (via NVIDIA)
  • Mistral Small 4 119B (via NVIDIA)
bash
# Use free models only
/model free

Free models are capable

The free NVIDIA models are surprisingly strong for most coding tasks. Try them before spending credits — they may be all you need.

Switching Models

Switch to a specific model mid-conversation with the /model command:

bash
# Switch to a specific model
/model claude-sonnet-4.6
/model gpt-5
/model gemini-2.5-pro

# Switch to a routing profile
/model auto
/model free

Model changes take effect immediately on the next message.

Prompt Caching

Anthropic models (Claude) support prompt caching, which reduces cost by approximately 75% for repeated context. This is enabled automatically — long conversations and system prompts are cached and reused across messages.

Caching is automatic

You don't need to configure anything. Franklin detects cacheable content and sends the appropriate cache headers to Anthropic's API automatically.