LLM Prices
2,200+ models. Real costs. No BS.
OpenCode Go — $10/mo
$60 of usage included. Real multiplier depends on model.
| Model | Requests/mo | Value | Multiplier | Cost/Req | Context | Best For |
|---|---|---|---|---|---|---|
| Muse Spark 1.2 | 226,600 | $78 | 7.8x | $0.0003 | 1M | Volume tasks |
| MiMo-V2.5 | 150,400 | $60 | 6.0x | $0.0004 | 1M | Bulk coding, agents |
| DeepSeek V4 Flash | 37,800 | $22 | 2.2x | $0.0006 | 1M | Fast inference |
| Qwen3.7 Plus | 21,600 | $22 | 2.2x | $0.0010 | 256K | General coding |
| Hy3 | 21,500 | $10 | 1.0x | $0.0005 | 1M | Reasoning |
| MiniMax M3 | 16,000 | $13 | 1.3x | $0.0008 | 1M | General |
| Kimi K2.7 Code | 6,750 | $15 | 1.5x | $0.0022 | 128K | Code generation |
| GPT 5.6 Luna | 10,250 | $7 | 0.7x | $0.0007 | 272K | Quality code |
| DeepSeek V4 Pro | 5,200 | $7 | 0.7x | $0.0013 | 1M | Pro reasoning |
| Grok 4.5 | 600 | $2 | 0.2x | $0.0036 | 500K | Frontier reasoning |
Cheapest API Models
Pay-per-token. Blended = input + 3×output per 1M tokens.
| Model | Provider | Blended/1M | Input/1M | Output/1M | Context |
|---|---|---|---|---|---|
| Llama 3.3 70B | Groq | FREE | $0 | $0 | 128K |
| Llama 3.1 8B | Cerebras | FREE | $0 | $0 | 8K |
| Gemini 2.5 Flash | FREE | $0 | $0 | 1M | |
| DeepSeek V3.2 | DeepSeek | $3.58 | $0.28 | $1.10 | 128K |
| Llama 3.3 70B | Together AI | $3.52 | $0.88 | $0.88 | 128K |
| DeepSeek R1 | OpenRouter | $7.12 | $0.55 | $2.19 | 128K |
| Claude Opus 4.6 | Anthropic | $80.00 | $5.00 | $25.00 | 200K |
| GPT 5.5 | OpenAI | $32.50 | $2.50 | $10.00 | 128K |
Free Tiers
No credit card. Rate limits apply.
| Provider | Model | Context | Rate Limit |
|---|---|---|---|
| Groq | Llama 3.3 70B | 128K | 30 RPM |
| Cerebras | Llama 3.1 8B | 8K | 30 RPM |
| Google Gemini | Gemini 2.5 Flash | 1M | 10 RPM |
| OpenRouter | ~30 free models | varies | 20 RPM |
| GitHub Models | 100+ models | varies | 15 RPM |
| xAI | Grok | 128K | $25 credits |
When to Use What
Bulk Coding
MiMo-V2.5 — 150K req/mo, 6x value, 1M context. Tests, refactors, mechanical work.
Reasoning
Hy3 — Thinks before acting. Architecture, debugging, complex tasks.
Fast Inference
Groq free — 30 RPM, LPU hardware. Quick questions, chat.
Cheapest API
DeepSeek V3.2 — $3.58/1M blended. Pay-per-token, no subscription.
Quality Code
GPT 5.6 Luna — Best code quality, 272K context. Production code.
Free
Groq/Cerebras — No credit card. Light usage, prototyping.