Models
Models
What each provider charges to run an AI model, per million tokens, beside its context window, tool calling and published benchmark scores.
114 of 241 models priced across 37 providers · as of 26 Oct 2026
241 models
| Model | Licence | Context | Providers | Lowest price per 1M tokens |
|---|---|---|---|---|
| A.X K2 SK Telecom | Open weights | 262k | — | — |
| Apertus 70B Instruct Swiss AI Initiative | Open weights | 65.5k | — | — |
| Apertus 8B Instruct Swiss AI Initiative | Open weights | 65.5k | — | — |
| Apertus v1.5 70B / 8B Swiss AI Initiative | Open weights | 262k | — | — |
| Beam 1.0 501B-A23B Reflection AI | Open weights | 1.05M | 2 | $0.25 in · $1.00 out |
| Claude 4.5 Haiku Anthropic | Proprietary | 200k | — | — |
| Claude Fable 5 Anthropic | Proprietary | 1M | 6 | $10.00 in · $50.00 out |
| Claude Fable 5 (with fallback) Anthropic | Proprietary | 1M | — | $10.00 in · $50.00 out |
| Claude Fable 5.1 Anthropic | Proprietary | 1M | 4 | $10.00 in · $50.00 out |
| Claude Haiku 5.5 Anthropic | Proprietary | 1M | 4 | $0.10 in · $0.50 out |
| Claude Mythos 5.1 Anthropic | Proprietary | 1M | 2 | $10.00 in · $50.00 out |
| Claude Opus 4.7 Anthropic | Proprietary | 1M | 7 | $5.00 in · $25.00 out |
| Claude Opus 4.8 Anthropic | Proprietary | 1M | 6 | $5.00 in · $25.00 out |
| Claude Opus 5 Anthropic | Proprietary | 1M | 6 | $5.00 in · $25.00 out |
| Claude Opus 5.5 Anthropic | Proprietary | 1M | 4 | $4.00 in · $20.00 out |
| Claude Sonnet 4.6 Anthropic | Proprietary | 1M | 7 | $3.00 in · $15.00 out |
| Claude Sonnet 4.6 (32k) Anthropic | Proprietary | 1M | — | $3.00 in · $15.00 out |
| Claude Sonnet 5 Anthropic | Proprietary | 1M | 6 | $2.00 in · $10.00 out |
| Claude Sonnet 5.5 Anthropic | Proprietary | 1M | 4 | $2.00 in · $10.00 out |
| Command A+ Cohere | Open weights | — | 1 | $0 in · $0 out |
| DeepSeek V4 Flash DeepSeek | Open weights | 1.05M | 5 | $0.09 in · $0.18 out |
| DeepSeek V4 Flash 0731 DeepSeek | Open weights | 1.05M | 10 | $0.06 in · $0.18 out |
| DeepSeek V4 Flash Vision Exp DeepSeek | Open weights | 1.05M | 2 | $0.44 in · $1.32 out |
| DeepSeek V4 Pro DeepSeek | Open weights | 1.05M | 9 | $1.30 in · $2.60 out |
| DeepSeek V4 Pro (GA, 0813, Max) DeepSeek | Open weights | 1.05M | — | — |
| DeepSeek V4 Pro (GA, 0813) DeepSeek | Open weights | 1.05M | — | — |
| DeepSeek V4.1 Flash DeepSeek | Open weights | 1.05M | 10 | $0.20 in · $0.60 out |
| DeepSeek V4.1 Flash 1017 DeepSeek | Open weights | 1.05M | 10 | $0.20 in · $0.60 out |
| Devstral 2 Mistral | Open weights | — | — | — |
| Devstral Small 2 Mistral | Open weights | — | — | — |
| Doubao Seed Code ByteDance Seed | Proprietary | 256k | — | — |
| ERNIE 4.5 300B A47B Baidu | Open weights | 131k | — | — |
| ERNIE 5.0 Thinking Preview Baidu | Proprietary | 128k | — | — |
| Exaone 4.0 1.2B LG AI Research | Open weights | 65.5k | — | — |
| EXAONE 4.0 32B LG AI Research | Open weights | 131k | — | — |
| EXAONE 4.5 33B LG AI Research | Open weights | 262k | — | — |
| Falcon-H1R-7B TII UAE | Open weights | 262k | — | — |
| Gemini 2.5 Pro Google | Proprietary | 1.05M | 3 | $1.25 in · $10.00 out |
| Gemini 3 Deep Think Google | Proprietary | — | — | — |
| Gemini 3.1 Flash-Lite Google | Proprietary | 1.05M | 3 | $0.25 in · $1.50 out |
| Gemini 3.1 Pro Preview Google | Proprietary | 1.05M | 3 | $2.00 in · $12.00 out |
| Gemini 3.5 Flash Google | Proprietary | 1.05M | 3 | $1.50 in · $9.00 out |
| Gemini 3.5 Flash-Lite Google | Proprietary | 1.05M | 2 | $0.30 in · $2.50 out |
| Gemini 3.6 Flash Google | Proprietary | 1.05M | 2 | $1.50 in · $7.50 out |
| Gemini 3.7 Flash Google | Proprietary | 1.05M | 2 | $0.75 in · $3.75 out |
| Gemini 3.8 Flash Google | Proprietary | 1.05M | 2 | $1.50 in · $7.50 out |
| Gemini 4 Argon Google | Proprietary | 1.05M | 2 | $4.00 in · $20.00 out |
| Gemma 3 270M Google | Open weights | — | — | — |
| Gemma 4 12B Google | Open weights | — | — | — |
| Gemma 4 26B A4B Google | Open weights | — | — | — |
Prices are each provider's standard on-demand list price in US dollars per million tokens: default region, standard tier, base context tier. Benchmark scores come from each benchmark's maintainers, Epoch AI's independent runs, and figures the model makers published themselves (marked v); each score links to its source. Data as of 26 Oct 2026.
Sources and licences
- Epoch AI Benchmarking Hub (GPQA Diamond runs) — CC BY 4.0
- Terminal-Bench 2.1 leaderboard — Apache-2.0
- τ-bench leaderboard (Sierra) — MIT
- APEX-Agents leaderboard (Mercor) — CC BY 4.0
- CritPt v2 leaderboard — Apache-2.0
- MMMU leaderboard (MMMU-Pro values shared under the same licence) — CC BY-SA 4.0
- models.dev (context windows, licences, prices) — MIT
- LiteLLM model price list — MIT
- Vendor-reported scores: each links to the model maker's own page.
- Where open price lists disagreed, the provider's own pricing page decided.