LV
LLMversus

What's the Cheapest LLM for Coding?

Finding the cheapest LLM for coding requires balancing price with coding ability. Here are the most affordable coding-capable models, ranked by a weighted cost metric (60% input, 40% output):


  • Phi-4 (Microsoft): $0.070/M input, $0.140/M output. Coding ELO: 1130. Speed: 160 tok/s.
  • Gemini 2.0 Flash Lite (Google): $0.075/M input, $0.300/M output. Coding ELO: 1170. Speed: 180 tok/s.
  • Llama 4 Scout (Meta): $0.100/M input, $0.300/M output. Coding ELO: 1230. Speed: 110 tok/s.
  • Mistral Small (Mistral): $0.100/M input, $0.300/M output. Coding ELO: 1160. Speed: 120 tok/s.
  • Gemini 2.0 Flash (Google): $0.100/M input, $0.400/M output. Coding ELO: 1240. Speed: 160 tok/s.

    • For simple code generation and autocomplete, smaller models like GPT-4.1 Nano or Gemini 2.0 Flash Lite are extremely affordable. For complex multi-file refactoring and architecture decisions, investing in Claude Sonnet 4 or GPT-4.1 pays off in fewer iterations and better results.


    Cost-saving tips for coding workloads: Use prompt caching for system prompts with coding instructions. Batch non-urgent code reviews through batch APIs. Start with a cheaper model and only escalate to premium models for complex tasks.

    Related Tools

    Full Pricing TableCost CalculatorSpeed Comparison

    Related Questions

    How Much Does Claude API Cost?

    Complete breakdown of Claude API pricing across all models — Opus 4, Sonnet 4, and Haiku 4.

    ChatGPT vs Claude: Which Is Better?

    Head-to-head comparison of OpenAI and Anthropic models across pricing, benchmarks, and capabilities.

    Best LLM API for Production Use

    Which LLM API is best for production? Evaluating reliability, speed, pricing, and capabilities.

    LLM API Pricing Comparison — Complete Guide

    Complete guide to LLM API pricing in 2026 across all major providers.

    How to Reduce LLM API Costs

    Practical strategies for cutting LLM API spend: caching, batching, model routing, and more.

    Which LLM Has the Largest Context Window?

    Ranking of all LLMs by context window size, from 1M tokens down to 32K.

    Fastest LLM API — Speed Comparison

    Which LLM API is fastest? Comparing TTFT and throughput across all major providers.