Models
Compare what providers charge to run each AI model, per million tokens, beside its published benchmark scores. Every price is read from the provider's own site and carries the day it was checked.
114 models priced across 36 providers · prices checked 25 Oct – 18 Nov 2026
232 models · 104 on the chart · 10 known by name only, hidden · · showing the first 50 · sorted by most providers
Price against benchmark score
Lowest listed blended price per 1M tokens (log scale) against the best published score
What this shows: 104 models have both a listed price and a GPQA Diamond score; the line joins the 8 that no other plotted model beats on both price and score. 2 free listings are not plotted.
- Round: open weights
- Square: proprietary
- No outline: licence not listed
- Dashed: score reported by the model's maker
- No plotted model is both cheaper and higher-scoring
Each mark is a model, shown by its maker's logo, or by initials where we have none; in a crowded spot the logo is drawn smaller. Price is the lowest per-token list price; the score is the model's best published score across its settings. A higher setting can use more tokens per task, which a per-token price does not show. Tap a point for its details.
Model list
What the numbers mean
- Input and output
- The lowest input price and the lowest output price any listed provider or the maker's own API charges, per million tokens; the two can come from different listings.
- Blended
- Three input tokens to one output token, the rate the price sort uses.
- GPQA Diamond
- The model's best published score across its settings, from the benchmark's maintainers or an independent evaluator before the maker's own figure; v marks a score the maker reported.
- Context
- The context window: how many tokens the model can take in at once.
- Providers
- How many of the providers we track serve this model.
The lowest input price any listed provider or the maker's own API charges, per million tokens.per 1M | The lowest output price any listed provider or the maker's own API charges, per million tokens. It can come from a different listing than the lowest input price.per 1M | Three input tokens to one output token, the rate the price sort uses. The strip runs from the lowest to the highest provider; the diamond is the maker's own API.per 1M | The model's best published score across its settings, from the benchmark's maintainers or an independent evaluator before the maker's own figure; v marks a score the maker reported.best published | How many of the providers we track serve this model. | ||
|---|---|---|---|---|---|---|
GLM-5.2 Z.ai · Open weights | $0.19 in | $2.40 out | $1.16 | GPQA Diamond 89.5%max effort | 1.05M context | 17 providers |
gpt-oss-120b OpenAI · Open weights | $0.037 in | $0.17 out | $0.07 | GPQA Diamond 78.2%high effort | 131k context | 15 providers |
GLM-5.3 Z.ai · Open weights | $0.90 in | $4.00 out | $1.68 | GPQA Diamond 91.7%max effort | 1.05M context | 13 providers |
GLM-5.3-Flash Z.ai · Open weights | $0.15 in | $0.50 out | $0.24 | GPQA Diamond 91.2% | 1.05M context | 12 providers |
Kimi K3 Moonshot AI · Open weights | $2.70 in | $13.50 out | $5.40 | GPQA Diamond 93.5%max effort | 1.05M context | 12 providers |
DeepSeek V4 Flash 0731 DeepSeek · Open weights | $0.06 in | $0.18 out | $0.09 | GPQA Diamond 90.8%max effort | 1.05M context | 10 providers |
DeepSeek V4.1 Flash DeepSeek · Open weights | $0.20 in | $0.60 out | $0.30 | GPQA Diamond 89.8%max effort | 1.05M context | 10 providers |
DeepSeek V4.1 Flash 1017 DeepSeek · Open weights | $0.20 in | $0.60 out | $0.30 | GPQA Diamond 93.2%max effort | 1.05M context | 10 providers |
DeepSeek V4 Pro DeepSeek · Open weights | $1.30 in | $2.60 out | $1.62 | GPQA Diamond 90.5%high effort | 1.05M context | 9 providers |
Kimi K2.6 Kimi · Open weights | $0.75 in | $3.40 out | $1.44 | GPQA Diamond 91.1%thinking on | 262k context | 8 providers |
Kimi K2.7 Code Kimi · Open weights | $0.86 in | $3.80 out | $1.59 | GPQA Diamond 89.6% | 262k context | 8 providers |
MiniMax-M3 MiniMax · Open weights | $0.28 in | $1.10 out | $0.49 | GPQA Diamond 92.9% | 1M context | 8 providers |
Claude Opus 4.7 Anthropic · Proprietary | $5.00 in | $25.00 out | $10.00 | GPQA Diamond 91.4%max effort | 1M context | 7 providers |
Claude Sonnet 4.6 Anthropic · Proprietary | $3.00 in | $15.00 out | $6.00 | GPQA Diamond 87.5%max effort | 1M context | 7 providers |
Qwen3.8 27B Alibaba · Open weights | $0.20 in | $1.49 out | $0.78 | GPQA Diamond 90.5%xhigh effort | 262k context | 7 providers |
Claude Fable 5 Anthropic · Proprietary | $10.00 in | $50.00 out | $20.00 | GPQA Diamond 92.6%max effort, production safeguards on | 1M context | 6 providers |
Claude Opus 4.8 Anthropic · Proprietary | $5.00 in | $25.00 out | $10.00 | GPQA Diamond 92%max effort | 1M context | 6 providers |
Claude Opus 5 Anthropic · Proprietary | $5.00 in | $25.00 out | $10.00 | GPQA Diamond 93.7%xhigh effort | 1M context | 6 providers |
Claude Sonnet 5 Anthropic · Proprietary | $2.00 in | $10.00 out | $4.00 | GPQA Diamond 91.1%max effort | 1M context | 6 providers |
GPT-5.5 OpenAI · Proprietary | $5.00 in | $30.00 out | $11.25 | GPQA Diamond 93.5%xhigh effort | 1.05M context | 6 providers |
gpt-oss-20B OpenAI · Open weights | $0.03 in | $0.14 out | $0.058 | GPQA Diamond 68.8%high effort | 131k context | 6 providers |
MiniMax M3.1 Flash MiniMax · Proprietary | $0.15 in | $0.60 out | $0.26 | — | 1M context | 6 providers |
Qwen3.5 397B A17B Alibaba · Open weights | $0.45 in | $3.00 out | $1.09 | GPQA Diamond 89.3%thinking on | 262k context | 6 providers |
Qwen3.8 Max Alibaba · Proprietary | $1.65 in | $4.95 out | $2.48 | GPQA Diamond 92.8% | 1M context | 6 providers |
DeepSeek V4 Flash DeepSeek · Open weights | $0.09 in | $0.18 out | $0.11 | GPQA Diamond 89.4%max effort | 1.05M context | 5 providers |
GLM-5.1 Z AI · Open weights | $1.38 in | $4.40 out | $2.13 | GPQA Diamond 86.8%thinking on | 203k context | 5 providers |
GPT-5.6 Luna OpenAI · Proprietary | $0.20 in | $1.20 out | $0.45 | GPQA Diamond 91.1%max effort | 1.05M context | 5 providers |
GPT-5.6 Sol OpenAI · Proprietary | $4.00 in | $20.00 out | $8.00 | GPQA Diamond 94.1%max effort | 1.05M context | 5 providers |
GPT-5.6 Terra OpenAI · Proprietary | $2.00 in | $12.00 out | $4.50 | GPQA Diamond 92.5%max effort | 1.05M context | 5 providers |
Qwen3.6 35B A3B Alibaba · Open weights | $0.10 in | $0.95 out | $0.31 | GPQA Diamond 84.1%thinking on | 262k context | 5 providers |
Qwen3.7 Max Alibaba · Proprietary | $1.25 in | $3.75 out | $1.88 | GPQA Diamond 92.3% | 1M context | 5 providers |
Qwen3.8 Flash Alibaba · Proprietary | $0.11 in | $0.38 out | $0.18 | GPQA Diamond 91.7%v, reported by the maker | 1M context | 5 providers |
Claude Fable 5.1 Anthropic · Proprietary | $10.00 in | $50.00 out | $20.00 | GPQA Diamond 93.7%max effort, production safeguards on | 1M context | 4 providers |
Claude Haiku 5.5 Anthropic · Proprietary | $0.10 in | $0.50 out | $0.20 | GPQA Diamond 89.6%max effort | 1M context | 4 providers |
Claude Opus 5.5 Anthropic · Proprietary | $4.00 in | $20.00 out | $8.00 | GPQA Diamond 90.6%max effort | 1M context | 4 providers |
Claude Sonnet 5.5 Anthropic · Proprietary | $2.00 in | $10.00 out | $4.00 | GPQA Diamond 95.6%max effort | 1M context | 4 providers |
Grok 4.3 xAI · Proprietary | $1.25 in | $2.50 out | $1.56 | GPQA Diamond 90.1%high effort | 1M context | 4 providers |
Grok 4.6 xAI · Proprietary | $2.00 in | $6.00 out | $3.00 | GPQA Diamond 94.9%high effort | 500k context | 4 providers |
Qwen3.5 35B A3B Alibaba · Open weights | $0.14 in | $1.00 out | $0.35 | GPQA Diamond 84.5%thinking on | 262k context | 4 providers |
Qwen3.6 27B Alibaba · Open weights | $0.30 in | $3.20 out | $1.02 | GPQA Diamond 84.2%thinking on | 262k context | 4 providers |
Gemini 2.5 Pro Google · Proprietary | $1.25 in | $10.00 out | $3.44 | GPQA Diamond 84.4% | 1.05M context | 3 providers |
Gemini 3.1 Flash-Lite Google · Proprietary | $0.25 in | $1.50 out | $0.56 | GPQA Diamond 82.2% | 1.05M context | 3 providers |
Gemini 3.1 Pro Preview Google · Proprietary | $2.00 in | $12.00 out | $4.50 | GPQA Diamond 94.1% | 1.05M context | 3 providers |
Gemini 3.5 Flash Google · Proprietary | $1.50 in | $9.00 out | $3.38 | GPQA Diamond 92.2%high effort | 1.05M context | 3 providers |
GPT-5.4 mini OpenAI · Proprietary | $0.75 in | $4.50 out | $1.69 | GPQA Diamond 87.5%xhigh effort | 400k context | 3 providers |
GPT-5.4 nano OpenAI · Proprietary | $0.20 in | $1.25 out | $0.46 | GPQA Diamond 81.7%xhigh effort | 400k context | 3 providers |
GPT-6 Astra OpenAI · Proprietary | $10.00 in | $50.00 out | $20.00 | GPQA Diamond 96.3%xhigh effort | 1.05M context | 3 providers |
GPT-6 Luna OpenAI · Proprietary | $0.10 in | $0.50 out | $0.20 | GPQA Diamond 90.5%max effort | 1.05M context | 3 providers |
GPT-6 Sol OpenAI · Proprietary | $2.00 in | $10.00 out | $4.00 | GPQA Diamond 94.3%max effort | 1.05M context | 3 providers |
GPT-6.1 Sol OpenAI · Proprietary | $2.00 in | $10.00 out | $4.00 | GPQA Diamond 95.4%max effort | 1.05M context | 3 providers |
QAI lists these prices; it does not serve or resell these models. Prices are each provider's standard on-demand list price in US dollars per million tokens (default region, standard tier, peak hours, base context tier), checked on the date shown against the provider's own published prices or, where it has no public price page, its listing on OpenRouter. Most benchmark scores come from Artificial Analysis; Epoch AI (GPQA Diamond), MathArena (AIME 2026) and the APEX-Agents and CritPt leaderboards add the rest, and where none of them has a model, or none ran it at full capability, a figure its maker published may stand in (marked v); in some benchmarks it was run the maker's own way, which its note names. Each benchmark follows one test setup, the one used by the source that covers the most models, and leaves results from other setups out. Each score is the best its source published across the model's settings, names any non-default setting, and links to its source.
An AIME 2026 score noted “released after the exam” comes from a model released after the exam, so the problems may have been in its training data. GDPval-AA is a rating against other models (Elo), not a percentage. Data as of 26 Oct 2026.
Sources and licences
- Artificial Analysis (its own evaluations and its runs of public benchmarks)
- Epoch AI Benchmarking Hub (GPQA Diamond runs) — CC BY 4.0
- MathArena (AIME 2026 runs)
- APEX-Agents leaderboard (Mercor) — CC BY 4.0
- CritPt v2 leaderboard — Apache-2.0
- 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 these price lists and a provider's own pricing page disagreed, the provider's page won.