Pricing guide
Cheapest AI Models That Still Hold Up
Cheap is only useful if the model is still competent. These are the low-cost options that still deserve production trials.
| Model | Input | Output | Total price signal | Best for |
|---|---|---|---|---|
| GPT-OSS-120B OpenAI | $0 | $0 | $0.00 | Self-hosted |
| DeepSeek V4 Pro DeepSeek | $0.435 | $0.87 | $1.30 | Budget coding |
| MiniMax M2.5 MiniMax | $0.3 | $1.2 | $1.50 | Real-world productivity |
| GLM-5 Zhipu AI | $0.5 | $2 | $2.50 | Bilingual (CN/EN) |
| Kimi K2.5 Moonshot AI | $0.6 | $2.5 | $3.10 | Visual coding |
| Grok 4.3 xAI | $1.25 | $2.5 | $3.75 | Long context |
| Qwen3 Max 2026-01-23 Alibaba | $1.2 | $6 | $7.20 | Multilingual |
| Mistral Medium 3.5 Mistral | $2 | $6 | $8.00 | European compliance |
| Llama 4 Maverick Meta | $2 | $8 | $10.00 | Open weights |
| Gemini 3.5 Flash | $1.5 | $9 | $10.50 | Fast multimodal agents |
| Gemini 3.1 Pro Preview | $2 | $12 | $14.00 | Multimodal tasks |
| GPT-5.2-Codex OpenAI | $1.75 | $14 | $15.75 | Coding-focused tasks |
| GPT-5.2 OpenAI | $1.75 | $14 | $15.75 | General-purpose |
| GPT-5.4 OpenAI | $2.5 | $15 | $17.50 | Coding |
| Claude Sonnet 4.6 Anthropic | $3 | $15 | $18.00 | Balanced performance |
| Claude Opus 4.8 Anthropic | $5 | $25 | $30.00 | Complex reasoning |
| GPT-5.5 OpenAI | $5 | $30 | $35.00 | Complex reasoning |
Best ultra-budget option
DeepSeek V4 Pro combines low listed API pricing, 1M context, and hybrid thinking modes for cost-sensitive production trials.
Best value frontier model
MiniMax M2.5 is the best balance of serious quality and low token cost.
Best bilingual value
GLM-5 is still one of the strongest value models if Chinese/English matters.
When cheap is the right choice
- High-volume automations
- Background summarization and extraction
- Internal tooling where cost beats polish
- Early-stage startups watching burn