FastMetal

Rankings by use case

The best LLM APIs you can call today, by use case

Only models FastMetal serves, ranked per use case. Positions come from the public arena's ELO; prices are per 1M tokens. As of October 2026.

Coding

  1. 1Claude Fable 5$52.75
  2. 2Claude Opus 5.5$21.10
  3. 3Claude Opus 5$26.38
  4. 4Claude Opus 4.6$26.77
  5. 5GPT-6.1 Sol$10.55

Japanese

  1. 1Claude Fable 5.1$52.75
  2. 2Claude Fable 5$52.75
  3. 3Gemini 3.7 Flash$3.96
  4. 4Claude Opus 5$26.38
  5. 5Kimi K3$15.83

Overall

  1. 1Claude Opus 5.5$21.10
  2. 2Claude Opus 5$26.38
  3. 3Claude Fable 5$52.75
  4. 4Claude Fable 5.1$52.75
  5. 5Claude Opus 4.6$26.77

Math

  1. 1Claude Fable 5$52.75
  2. 2Claude Fable 5.1$52.75
  3. 3Gemini 3.8 Flash$3.96
  4. 4Claude Opus 5$26.38
  5. 5Claude Sonnet 5.5$10.55

Writing

  1. 1Claude Opus 5.5$21.10
  2. 2Claude Opus 5$26.38
  3. 3Claude Fable 5$52.75
  4. 4Gemini 3.7 Flash$3.96
  5. 5Gemini 3.8 Flash$3.96

Instruction following

  1. 1Claude Opus 5.5$21.10
  2. 2Claude Opus 5$26.38
  3. 3Claude Fable 5$52.75
  4. 4Claude Opus 4.6$26.77
  5. 5Claude Fable 5.1$52.75

Hard prompts

  1. 1Claude Opus 5.5$21.10
  2. 2Claude Opus 5$26.38
  3. 3Claude Fable 5$52.75
  4. 4Claude Opus 4.6$26.77
  5. 5Claude Fable 5.1$52.75

Multi-turn chat

  1. 1Claude Fable 5$52.75
  2. 2Claude Opus 4.6$26.77
  3. 3Muse Spark 1.2$4.68
  4. 4Gemini 3.8 Flash$3.96
  5. 5Kimi K3$15.83

Vision

  1. 1Claude Fable 5$52.75
  2. 2Qwen3.8 Max$6.33
  3. 3Gemini 3.7 Flash$3.96
  4. 4Claude Opus 4.6$26.77
  5. 5GPT-6.1 Sol$10.55

Price

  1. 1Llama 3.1 8B Instruct$0.0422
  2. 2Voxtral Mini 3B$0.0528
  3. 3gpt-oss-20b$0.1477
  4. 4Mercury 2.5$0.165
  5. 5Ling 3.0 Flash VL$0.1899

Long context

  1. 1GLM 5.3 Flash$0.5275
  2. 2GPT-5.6 Luna$1.27
  3. 3GPT-5.6 Sol$11.00
  4. 4GPT-5.6 Terra$12.66
  5. 5GPT-6 Astra$52.75

Price is per 1M output tokens