Models included in estimatesQuick token calculator →
Anthropic
Claude Haiku 5.5*Low-cost$0.10 / $0.50
Claude Sonnet 5.5Mid-tier$2.00 / $10.00
Claude Opus 5.5Premium$4.00 / $20.00
Claude Fable 5.1Premium$10.00 / $50.00
OpenAI
GPT-6 LunaLow-cost$0.10 / $0.50
GPT-6.1 SolMid-tier$2.00 / $10.00
GPT-6 AstraPremium$10.00 / $50.00
Google
Gemini 3.5 Flash-LiteLow-cost$0.30 / $2.50
Gemini 3.8 Flash*Mid-tier$0.75 / $3.75
Gemini 4 Argon*Premium$2.00 / $10.00
xAI
Grok 4.7*Mid-tier$2.00 / $6.00
Open-weight
Llama 3.3 70B*Mid-tier$1.04 / $1.04
Mistral Large 3*Mid-tier$0.50 / $1.50
Muse Spark Open*Mid-tier$1.25 / $4.25
Other API
DeepSeek V4.1 Flash*Low-cost$0.30 / $1.20
DeepSeek V4 Pro*Mid-tier$1.32 / $3.96
Qwen3.8 Flash*Low-cost$0.15 / $0.47
Qwen3.8 Max*Mid-tier$2.00 / $6.00
Price notes. These prices can change or depend on how you use the model. Confirm current pricing with the provider before you build.
  • Claude Haiku 5.5: Price rises to $0.50 in / $2.50 out per 1M tokens for prompts over 100,000 tokens.
  • Gemini 3.8 Flash: Promotional price through Dec 31, 2026. From Jan 1, 2027 it is $1.50 in / $7.50 out per 1M tokens.
  • Gemini 4 Argon: Promotional price through Nov 30, 2026. From Dec 1, 2026 it is $4.00 in / $20.00 out per 1M tokens.
  • Grok 4.7: Prompts of 200k tokens or more are billed at $4.00 in / $12.00 out per 1M tokens, for the whole request.
  • Llama 3.3 70B: Open-weight model: priced at Together AI's serverless rate. Other hosts charge differently, and self-hosting costs depend on your hardware.
  • DeepSeek V4.1 Flash: Peak-hours rate shown. Off-peak is half price, and DeepSeek sets the peak windows.
  • DeepSeek V4 Pro: Peak-hours rate shown. Off-peak is half price, and DeepSeek sets the peak windows.
  • Qwen3.8 Flash: International (Singapore) pricing; other regions are priced differently.
  • Qwen3.8 Max: International (Singapore) pricing; other regions are priced differently.
  • Mistral Large 3: Open-weight model: what you pay depends on where you run it. This is a hosted-API estimate; other hosts charge differently, and self-hosting costs depend on your hardware.
  • Muse Spark Open: Open-weight model: priced at Together AI's serverless rate. Other hosts charge differently, and self-hosting costs depend on your hardware.