Cheapest good LLM for agentic work
- Ling 3.0 Flash leads
- $0.063
- models qualify
- 44
Output $/Mtok · read October 21
Cheapest output, $ per million tokens
- Mistral Nemo $0.03
- Gemma 3 4B $0.05
- S Llama 3 8B Lunaris $0.05
- Ling 3.0 Flash $0.063
- Mistral Small 3 $0.08
- Llama 3.1 8B Instruct $0.08
- gpt-oss-20b $0.09
- N Nex-N2.5-Mini $0.10
As of October 21, 2026, inclusionAI Ling 3.0 Flash is the cheapest model at or above the median AA Agentic Index result, at $0.063 per million output tokens, ahead of inclusionAI Ling 3.0 Flash VL ($0.18) and DeepSeek V4 Flash 0423 ($0.182).
This list narrows the best LLM for agents ranking.
| Rank | Model | Output $/Mtok | Coding evidence | AA Agentic Index | Coding index |
|---|---|---|---|---|---|
| 1 | Ling 3.0 Flash inclusionAI | $0.063 | 19.3 | Not benchmarked | |
| 2 | Ling 3.0 Flash VL inclusionAI | $0.18 | 28.7 | Not benchmarked | |
| 3 | DeepSeek V4 Flash 0423 DeepSeek | $0.182 | 26.3 | 54.6 | |
| 4 | GLM 5.3 Flash Z.ai | $0.50 | 50.9 | 71.0 | |
| 5 | Hy3 preview Tencent | $0.60 | 24.1 | 22.4 | |
| 6 | MiMo-V2.5-Pro Xiaomi | $0.609 | 21.3 | 40.0 | |
| 7 | DeepSeek V4 Flash Vision Exp DeepSeek | $0.6468 | 47.5 | 37.6 | |
| 8 | GPT-5.6 Luna OpenAI | $1.20 | 42.1 | 61.5 | |
| 9 | MiniMax M3 MiniMax | $1.20 | 29.5 | 57.8 | |
| 10 | T Inkling Small Thinking Machines | $1.20 | 23.5 | 29.9 | |
| 11 | DeepSeek V4 Flash 0731 DeepSeek | $1.32 | 41.0 | Not benchmarked | |
| 12 | Qwen3.8 27B Qwen | $1.35 | 45.8 | 73.6 | |
| 13 | GLM 5.3 Z.ai | $1.40 | 53.1 | 73.1 | |
| 14 | DeepSeek V4 Pro 0423 DeepSeek | $1.9105 | 26.3 | 35.3 | |
| 15 | DeepSeek V4 Pro 0813 DeepSeek | $1.98 | 41.3 | 54.6 | |
| 16 | Qwen3.6 27B Qwen | $2.00 | 18.5 | Not benchmarked | |
| 17 | Nemotron 3 Ultra NVIDIA | $2.40 | 20.1 | Not benchmarked | |
| 18 | Kimi K2.7 Code Moonshot AI | $3.35 | 21.0 | 39.1 | |
| 19 | Kimi K2.6 Moonshot AI | $3.41 | 20.5 | 44.2 | |
| 20 | Gemini 3.6 Flash Google | $3.75 | 29.0 | 48.3 |
How this list is built
Generated from the dataset on every refresh. Nobody picks the order by hand.
Models whose AA Agentic Index result is at or above 18.5, the median of every model on that board, ordered by output price, cheapest first. Free listings are left out: they are a tariff, not a price.
- OpenRouter, fetched
- Artificial Analysis (via OpenRouter), fetched