K-EXAONE 236B — API Pricing & Benchmarks
K-EXAONE 236B is LG AI Research's flagship reasoning model. The API costs $0.20 per 1M input tokens and $0.80 per 1M output tokens ($0.35 blended at a 3:1 ratio), with cached input at $0.10. It supports a 262K-token context window. Pricing was cut on 2026-07-18 (previously $0.60/$1.00 per 1M tokens).
Explore provider routes and additional benchmark sources for K-EXAONE 236B
| Provider | LG AI Research |
| Input price / 1M tokens | $0.20 |
| Output price / 1M tokens | $0.80 |
| Cached input / 1M tokens | $0.10 |
| Blended price (3:1) | $0.35 |
| Context window | 262K |
| Max output tokens | — |
| Knowledge cutoff | 2024-12 |
| License | open weights |
| Input modalities | T |
| Output speed (tok/s) | — |
| Released | — |
Where K-EXAONE 236B fits
On a blended 3:1 basis K-EXAONE 236B costs $0.35 per 1M tokens, cheaper than 88% of the 80 priced models in this index. As LG AI Research's flagship it is aimed at the hardest workloads — complex reasoning, agentic coding and multi-step tool use — where output quality dominates cost. The weights are open, so it can be self-hosted for data-residency or cost reasons instead of consumed via API. Cached input is priced at $0.10 (50% below standard input), which rewards prompt structures with long stable prefixes — see our caching guide.
Price history
| Period | Input $/1M | Output $/1M | Cached $/1M |
|---|---|---|---|
| 2026-07-18 → 2026-07-18 | $0.60 | $1.00 | — |
| 2026-07-18 → current | $0.20 | $0.80 | $0.10 |
List-price changes recorded by this site. The full dataset is public: price-history.json (CC BY 4.0).
Alternatives at a similar price
- GLM-4.5-Air (Zhipu AI (Z.ai)) — $0.43/1M blended, AA Index 16.7
- Codestral 25.08 (Mistral AI) — $0.45/1M blended
- GPT-5.6 Luna (OpenAI) — $0.45/1M blended, AA Index 52.3
- GPT-5.4 nano (OpenAI) — $0.46/1M blended, AA Index 39.7
About LG AI Research
LG AI Research publishes the K-EXAONE models as open weights, combining Korean-language strength with pricing among the cheapest of any flagship-class model tracked here.